Fact-checked by Grok 2 weeks ago

Mapillary

Mapillary is a crowdsourced street-level that enables users worldwide to capture, upload, and process geotagged photos and videos using various cameras, including smartphones and action cameras, to build comprehensive visual maps powered by algorithms. Founded in 2013 by Jan Erik Solem, Johan Gyllenspetz, and Peter Neubauer in , , the connects images across time and space to generate interactive, immersive views of urban and rural environments. The service has amassed over 2 billion images (as of 2024) contributed by a global network of volunteers spanning 190 countries, facilitating applications in , transportation assessment, , and . In June 2020, Mapillary was acquired by (now , Inc.) for an undisclosed amount, integrating its technology to enhance Meta's capabilities—including AI-driven processing—while maintaining commitments to initiatives like . Key features include automated data extraction—such as for road signs, lanes, and text—powered by ; for example, a 2018 partnership with utilized Rekognition for image processing. Mapillary's tools support both individual contributors and organizations, such as the International Red Cross, which uses the platform to map vulnerable areas for . By democratizing street-level , it complements services like but emphasizes open collaboration and scalability through .

History

Founding and Early Development

Mapillary was founded in September 2013 in , , by Jan Erik Solem, Johan Gyllenspetz, Peter Neubauer, and Yubin Kuang. The company's core vision centered on leveraging to enable scalable mapping through crowdsourced contributions of geotagged photographs, aiming to create an open, collaborative image-based representation of the world. This approach sought to democratize street-level imagery by allowing individuals and organizations worldwide to capture and share photos using everyday devices like smartphones, fostering frequent updates and global coverage beyond traditional vehicle-based systems. The platform's initial launch occurred in November 2013 with the release of an early beta app for , enabling users to capture and upload sequences of geotagged images while moving. This app marked the first public access to Mapillary's open-source sharing model, where photos were automatically processed and stitched into navigable street-level views. To broaden accessibility, an app followed in January 2014, expanding the contributor base and accelerating image collection across diverse mobile platforms. Early adoption grew rapidly, with the platform amassing 900,000 images within the first 14 weeks after the launch, reflecting strong initial engagement from thousands of users. By mid-, Mapillary had established a foundation of several thousand active contributors, laying the groundwork for its expansion as a crowdsourced mapping resource. A key enhancement came on September 10, , when support for uploading panoramic and spherical photos was added, allowing for immersive 360-degree captures that enriched the platform's visual depth and utility.

Funding and Growth Milestones

Mapillary secured its initial seed funding of $1.5 million in January 2015, led by with participation from Playfair Capital, Wellington Financial, and LDV Capital, to support early platform development and community expansion. In March 2016, the company raised $8 million in a led by Atomico, with additional investments from , LDV Capital, and Playfair Capital, enabling hires in engineering and to scale image processing capabilities. The Series B round followed in April 2018, bringing in $15 million led by Ventures, alongside Catalyst Fund, NavInfo, and previous investors, which funded advancements in global data collection and partnerships for automotive applications. These investments fueled significant dataset growth, with Mapillary surpassing 1 billion images by late 2019 through crowdsourced contributions worldwide. The platform reached the milestone of 2 billion images in August 2023, reflecting accelerated uploads from diverse contributors including individuals, organizations, and mapping initiatives. By 2024, the dataset had grown beyond 2 billion images, underscoring the platform's scale as a key open-source resource for geospatial data. As of early 2025, the platform was on track to reach 3 billion images, driven by ongoing community contributions and grant programs. Mapillary marked its 10-year anniversary in October 2023, coinciding with reaching the 2 billion image milestone in August 2023, and emphasizing the role of its global community in driving contributions from over 200 countries via mobile apps, cameras, and integrations with projects like .

Acquisition by

In June 2020, , Inc. (now , Inc.) announced the acquisition of Mapillary, a Swedish-based street-level imagery platform, to bolster its geospatial mapping initiatives and compete with established services like and . The deal, whose financial terms were not disclosed, integrated Mapillary into 's broader efforts to enhance open mapping through crowdsourced and technologies. This move aligned Mapillary with Meta's goal of improving global map accessibility for both human users and AI-driven applications. Following the acquisition, Mapillary underwent significant policy shifts to expand its reach, including making all platform imagery and data freely available for commercial use starting immediately in June 2020, whereas previously it had been restricted to non-commercial purposes under an open license. This change democratized access for developers and businesses, fostering greater integration with Meta's resources to advance processing for mapping tasks. Mapillary's operations continued independently under Meta's oversight, leveraging the parent company's computational infrastructure to scale imagery analysis and automate feature extraction. As of 2025, Mapillary remains actively maintained as part of ' geospatial portfolio, with ongoing platform enhancements ensuring its role in collaborative mapping. In October 2025, Meta released version 6.8.0 of the Mapillary app, introducing support for importing images captured by third-party camera applications to streamline user contributions. These updates reflect sustained investment in the platform's usability and integration with Meta's ecosystem, supporting broader applications in and .

Platform Features

Image Capture and Upload

Users contribute street-level images to Mapillary through various capture modes designed to facilitate comprehensive coverage. These include walking, where individuals capture imagery on foot; riding, typically via for stable, low-speed sequences; and driving, suitable for higher speeds up to 100 km/h to cover longer routes efficiently. Additionally, panorama mode employs 360-degree cameras to provide full environmental views in a single capture. All modes require , which embeds GPS coordinates and orientation data into each image to enable accurate on the platform; mobile apps handle this automatically, while desktop uploads necessitate pre-embedded metadata. Mapillary supports image capture and upload primarily through dedicated mobile applications for and devices, allowing users to record sequences directly within the or import existing photos. The , updated in October 2025 with version 6.8.0, introduced support for importing images captured by other camera apps or devices directly from the home view, enhancing flexibility for users relying on third-party tools. Similarly, the enables manual or automatic photo capture with built-in GPS tagging and editing before upload, supporting both and mobile data connections for seamless contribution. These apps ensure compatibility with a wide range of smartphones, though optimal results come from devices with high-resolution sensors. The apps also support video recording and upload, which is processed into image sequences with embedded GPS data. To maintain high-quality, mappable sequences, Mapillary provides specific guidelines for photo capture. Images should prioritize clarity with good lighting, focus, and minimal distortion, using format for uploads. For spacing, are recommended to be taken approximately 5 apart, achieved by distance-based capture ( in apps) or every 2 seconds at a steady pace, to ensure smooth blending and overlap for . Coverage guidelines emphasize systematic paths, such as imaging all sides of streets and avoiding gaps, to build dense, interconnected sequences that support effective street-level . Adhering to these practices helps contributors generate reliable data sequences that integrate well into the platform's global imagery collection.

Computer Vision Processing

Mapillary employs techniques to automatically process crowdsourced street-level images, transforming them into structured map data suitable for geospatial applications. The core begins with sequence reconstruction, followed by and semantic segmentation, enabling the extraction of features like road geometries and signage. This backend automation ensures scalability for deriving accurate, vectorized representations of urban environments from diverse image contributions. A foundational component is (SfM), which reconstructs scenes from 2D image sequences by estimating camera positions and generating sparse point clouds. Mapillary utilizes its open-source OpenSfM library for this purpose, aligning images based on feature matches to stitch them into coherent, traversable paths that mimic street-level navigation. This process creates smooth transitions between images and supports the geolocation of content without relying solely on GPS metadata, enhancing accuracy in areas with sparse satellite coverage. Key advancements in OpenSfM, such as incremental reconstruction algorithms, allow efficient handling of large-scale sequences by iteratively building models starting from initial image pairs. Object detection identifies discrete elements within images, such as traffic signs and lane markings, using deep learning models trained on diverse global datasets. For instance, Mapillary's system recognizes over 1,500 traffic sign types across more than 100 countries, triangulating their positions from detections in multiple overlapping images to produce precise points for . Lane detection similarly outlines road boundaries, contributing to vectorized lane graphs that inform and autonomous systems. These detections are filtered for , requiring confirmation across at least two or three images to generate reliable map features. Semantic segmentation provides pixel-level classification, assigning labels from a palette of 124 object classes—such as , buildings, and —to entire images, which aids in broader understanding and feature delineation. Models like those benchmarked on the Mapillary Vistas achieve state-of-the-art performance on scene segmentation tasks, including the Cityscapes challenge, enabling the conversion of raster image data into scalable vector representations for GIS integration. This technique supports applications in by generating detailed land-use maps from processed imagery. Since its inception in , Mapillary's processing has evolved from basic SfM implementations to sophisticated AI-driven pipelines, with OpenSfM open-sourced in 2014 and subsequent enhancements like denser point clouds in 2016 improving reconstruction quality. Following the 2020 acquisition by , the technology integrated with Meta's AI resources, boosting scalability for handling billions of images and refining algorithms for combined aerial and street-level . These developments have enabled more robust, automated map updates at global scale.

Viewing and Editing Tools

Mapillary provides a web-based explorer that enables users to browse street-level imagery and derived map data interactively. The platform's web app allows navigation through global coverage by searching locations, zooming on a map interface, and viewing sequences of images captured along paths such as roads or trails. Users can access 360-degree panoramic views for immersive exploration, particularly useful for assessing environments in detail, and switch between perspectives to follow image sequences seamlessly. This explorer also supports filtering by object classes, such as traffic signs, to highlight AI-processed detections overlaid on the imagery. Editing functionalities empower contributors to refine imagery and map data directly within integrated tools. Users can tag objects by leveraging AI-generated detections, such as identifying street assets or road features, and incorporate these into edits for accuracy. Correcting sequences involves adjusting geolocation or through the desktop uploader, which processes and re-uploads modified image sets to align with . For broader contributions, the platform facilitates edits in collaborative environments, where users verify and update elements like points of interest based on viewed imagery. Integrations extend viewing and editing to professional geospatial software, notably . The add-in allows users to import Mapillary sequences into projects for advanced navigation, enabling digitization of features directly from street-level views and validation of asset existence. Similarly, widgets for Online and Web AppBuilder support embedding interactive imagery layers, facilitating edits to feature classes with real-time attribution updates. These tools integrate processed outputs like object detections to streamline workflows in mapping applications.

Datasets

Internal Dataset Scale

Mapillary's internal dataset began modestly following its launch in 2013, with the first street-level images uploaded that year. By mid-2014, the platform had accumulated approximately 900,000 images through early community contributions, marking rapid initial growth driven by user uploads via mobile apps and dashcams. This foundation expanded significantly over the subsequent years, reaching 500 million images by April 2019 and surpassing 1 billion by December of that year, fueled by ongoing crowdsourced uploads from volunteers worldwide. By early 2024, the dataset had grown to over 2 billion images, reflecting a decade of sustained community participation and technological advancements in image processing. The dataset's composition emphasizes diverse street-level imagery captured primarily through smartphone apps, action cameras, and vehicle-mounted devices, with community uploads continuing to add millions of images annually. This organic accumulation has prioritized breadth over depth in many regions, enabling the platform to serve as a global repository for map enhancement. As of , Mapillary's internal collection approaches 3 billion images, with projections indicating further expansion through persistent volunteer efforts. Geographically, the dataset exhibits uneven but progressively inclusive distribution, with denser coverage in urban and developed areas of , , and parts of , where over 170 million images were available in alone by early 2024. Crowdsourcing has been instrumental in extending reach to underrepresented regions, including rural and remote locales in , , and , where volunteer contributions help mitigate coverage gaps inherent to traditional mapping efforts. For instance, targeted community drives have boosted imagery in areas like the , enhancing data for disaster resilience and infrastructure monitoring. In 2025, integrations such as the with MapSwipe have accelerated volunteer-driven expansions by incorporating Mapillary's street-level into tasking platforms, allowing contributors to validate and map features like road conditions and environmental details in underserved areas. This , launched in March 2025 and expanded to web-based tools by mid-year, has facilitated more precise, community-led data collection, further scaling the dataset's utility for global ping initiatives.

Major External Contributions

In 2018, the Agency of Transportation became the first to upload its photologs to Mapillary, contributing approximately 5 million images captured over five years along state highways and local roads. These images provided detailed street-level coverage of 's rural areas and small towns, previously underrepresented in global mapping platforms. That same year, the added about 4.7 million street-level images to Mapillary, encompassing all state highways and enabling broader access to visual data for mapping and analysis. This upload enhanced coverage across Arizona's diverse terrain, including remote and arid regions where individual contributions are sparse. Other notable external contributions include partnerships with urban planning organizations, such as the 2017 collaboration with Mexico's IMPLAN , which integrated Mapillary for crowdsourced imagery in support of sustainable mobility and territorial planning initiatives. Through and provision, this effort generated datasets focused on key areas like the city's historical , drainage systems, and public spaces. Such governmental and organizational inputs have played a key role in addressing coverage gaps in areas with limited volunteer participation, particularly rural or under-mapped locales, thereby complementing Mapillary's community-driven growth to achieve more comprehensive global street-level imagery.

Released Research Datasets

Mapillary has released several publicly available datasets derived from its street-level imagery collection, specifically curated and annotated to support research in , autonomous driving, and geospatial AI. These datasets are provided under non-commercial licenses, such as Creative Commons Attribution-NonCommercial-ShareAlike (CC BY-NC-SA), to facilitate academic and research applications while restricting commercial use without permission. The flagship release is the Mapillary Vistas Dataset, launched in May 2017, which includes 25,000 high-resolution images sourced from diverse global locations and captured under varying weather, seasonal, and lighting conditions. This dataset features pixel-accurate semantic segmentation annotations across 66 object categories—such as vehicles, pedestrians, traffic signs, and road markings—along with instance-specific labels for 37 of those classes, enabling detailed scene understanding for applications like autonomous vehicle . Annotation guidelines emphasized human-verified, high-quality labeling to ensure consistency, with a focus on street-level diversity to address gaps in existing datasets like Cityscapes. Subsequent releases have expanded on specialized tasks. The Mapillary Traffic Sign Dataset, introduced in 2020, comprises over 100,000 images with bounding box annotations for more than 300 traffic sign classes, incorporating both manual and machine-assisted labeling to support global detection and fine-grained classification; guidelines prioritized hierarchical class structures to capture regional variations. Similarly, the Mapillary Street-Level Sequences (MSLS) Dataset, also released in 2020, offers 1.6 million images organized into short sequences with GPS metadata, designed for lifelong place recognition across urban and suburban environments, with annotations focusing on sequence-level ground truth for retrieval tasks. Other notable sets include the 2020 Mapillary Planet-Scale Depth Dataset, providing depth estimates for 750,000 images to train monocular depth estimation models, and more recent contributions like the Mapillary Metropolis Dataset for multi-modal 3D urban scene benchmarking and the Mapillary CrowdDriven Dataset for visual localization using human-annotated pose pairs. Across these, annotation protocols typically involve a combination of expert human review and automated pre-labeling to balance scale and accuracy. As of 2025, these datasets have seen extensive adoption in academic research, with the Vistas Dataset alone cited in over 1,300 peer-reviewed papers, influencing advancements in semantic segmentation models like Meta's Segment Anything (SAM) for panoptic tasks and fine-grained traffic sign validation frameworks. Usage extends to exploratory analyses, including studies on contributor patterns in Mapillary's crowdsourced data, which leverage dataset metadata to examine geographic coverage and annotation biases in global street imagery. For instance, recent works have utilized MSLS for sequence-based localization benchmarks and Traffic Sign data for cross-cultural sign detection, demonstrating their role in scaling AI training for real-world geospatial applications.

Licensing and Policies

Content Licensing

Mapillary has adopted the Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) license for all user-contributed images since April 29, 2014, allowing free use, sharing, and adaptation with attribution and share-alike requirements. This license applies to the platform's vast collection of street-level imagery, ensuring openness while protecting contributors' rights. For derived map data, such as extracted metadata like road signs or street numbers, Mapillary ensures compatibility with the (ODbL), facilitating integration with projects like . Following its acquisition by in June 2020, Mapillary has maintained its commitment to open principles, with all imagery now available free for both non-commercial and commercial use under the same CC BY-SA 4.0 terms. This evolution underscores the platform's ongoing dedication to accessible geospatial data while operating under Meta's oversight.

Usage and Attribution Requirements

Mapillary content, primarily street-level images and derived data, is licensed under the Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) framework, allowing broad access while imposing specific obligations on users. This license has permitted free use, including commercial applications, since its adoption in 2014, with no non-commercial restrictions applying to the core platform imagery. Users may view, download, distribute, and modify the content for various purposes, provided they adhere to the license terms. Attribution is mandatory for all uses of Mapillary content, particularly in derivative works such as edited images, maps, or integrated applications. Users must credit Mapillary explicitly, including a hyperlink to the original image page or the platform's homepage (www.mapillary.com), and provide a link to the CC BY-SA 4.0 license (https://creativecommons.org/licenses/by-sa/4.0/).[](https://www.mapillary.com/terms) For developer integrations or tools embedding Mapillary imagery, additional requirements apply, such as displaying the official Mapillary logo and ensuring links direct users to the source material without alteration. Failure to include these elements in publications, websites, or products constitutes a violation of the license. The policy distinguishes between direct use and modifications: while original content can be used commercially or non-commercially without fees post-2014, any derivatives—such as processed images, annotations, or combined datasets—must be shared under the same CC BY-SA 4.0 terms to maintain the share-alike condition. This ensures that enhancements or adaptations remain openly accessible, preventing lock-in of community-contributed data. Commercial entities, including those developing mapping services or models, benefit from this openness but must propagate the license in their outputs to comply. In 2025, Mapillary updated its tools to address imported images from third-party applications, enhancing platform flexibility while reinforcing user responsibilities. The app update in October introduced support for importing photos captured via external camera apps or devices, allowing seamless integration into Mapillary sequences. However, users importing such third-party content must verify and ensure compliance with any original source licenses, as Mapillary disclaims liability for external materials and treats them under the platform's general terms unless specified otherwise. These updates, effective alongside a revision in March, aim to streamline contributions without altering core attribution or share-alike mandates.

Applications and Impact

Mapping and Geospatial Integration

Mapillary plays a pivotal role in supporting (OSM) by enabling mappers to derive edits from its street-level imagery, particularly for updating roads, traffic signs, and other urban features. Contributors load Mapillary photos as custom overlays in OSM editors like and JOSM, allowing precise tracing and annotation based on visual sequences that capture sequential geographic progression. This integration has facilitated widespread image-derived contributions, with initiatives such as the 2025 CompleteTheMap US program encouraging new imagery uploads to enhance OSM coverage of under-mapped areas. The platform integrates seamlessly with leading geospatial tools, enhancing analysis workflows in professional environments. For users, Mapillary provides custom widgets for Web AppBuilder and direct support in and Online, enabling the incorporation of street-level sequences for viewing, editing, and layering over vector data to assess infrastructure changes. Similarly, supports Mapillary through basemap plugins that allow filtering by location, user, and timestamp, facilitating geospatial queries on features like road conditions or signage placement. These integrations empower analysts to combine Mapillary's visual data with GIS layers for tasks such as route optimization and . In , Mapillary's sequence data has been instrumental in rapid post-event assessments, as seen in 2025 collaborations for open mapping in the , where community-collected imagery aided in evaluating flood resilience and damage. Organizations like the Red Cross leverage these sequences alongside Humanitarian Team efforts to generate updated maps for emergency aid deployment. For , the platform supports detailed environmental analysis; a 2025 study utilized Mapillary imagery to measure pedestrian-level street greenery, informing city designs that prioritize walkable, green spaces over aerial views alone. Cities like have applied such data to inventory road assets, achieving near-complete coverage to guide upgrades and equitable development.

AI Training and Research

Mapillary's datasets, particularly the Vistas dataset, have been extensively utilized in training for tasks. Released in 2017, Vistas comprises 25,000 high-resolution street-level images annotated with pixel-accurate labels across 66 object categories, enabling the development of models for semantic segmentation, instance segmentation, and . Researchers have applied Vistas to train architectures such as variants for in autonomous driving scenarios, demonstrating improved generalization across diverse urban environments. For instance, studies have leveraged the dataset to benchmark fine-grained , revealing limitations in vision-language models and establishing baselines like DINOv2 for enhanced accuracy in self-driving applications. Academic research from 2019 to 2025 has increasingly drawn on Mapillary data to analyze contributor behaviors and derive streetscape metrics, advancing urban analytics. A 2020 exploratory study examined contribution patterns in Mapillary, identifying key factors influencing user participation, such as geographic coverage and temporal trends, to understand crowdsourced data quality and scalability. More recent work has focused on streetscape evaluation; for example, the "streetscape" R package, introduced in 2024, provides reproducible tools for processing Mapillary street view imagery, including functions for geospatial searches, semantic segmentation, and computation of metrics like the Green View Index (GVI) to quantify urban greenery and walkability. This package has supported studies on tree detection and perceptual assessments of street environments, facilitating broader research in sustainable urban planning. Mapillary has fostered collaborations with institutions to establish benchmarks in , promoting advancements in scene understanding and localization. Through initiatives like the Mapillary dataset, researchers have been invited to develop and test algorithms for city-scale 3D mapping, creating novel paradigms for evaluating in complex urban settings. Joint efforts, such as the 2019 ICCV with the COCO dataset organizers, have explored in street scenes, while datasets like Mapillary Street-Level Sequences (MSLS) provide benchmarks for lifelong place recognition tasks, made available specifically for use. These partnerships have contributed to high-impact outcomes, including victories in challenges like the 2018 ECCV Semantic Segmentation for autonomous .

Community and Commercial Uses

Mapillary has fostered numerous community initiatives that leverage its street-level imagery for volunteer-driven mapping efforts. In May 2025, Mapillary integrated with MapSwipe, a app developed by the Humanitarian OpenStreetMap Team (), enabling volunteers to access detailed street-level views for tasks such as identifying road surfaces, waste accumulation, and infrastructure in underserved regions worldwide. This partnership, supported by HeiGIT, has expanded MapSwipe's capabilities to include high-resolution validation of geospatial data, particularly in areas lacking traditional , thereby enhancing open mapping projects in developing countries. On the commercial front, Mapillary's imagery supports applications adopted by various municipalities. For instance, the City of utilized Mapillary to collect over 12.7 million images, initially using Trimble MX2 systems for comprehensive asset management, sharing the data across more than 30 city departments to inform infrastructure decisions. Similarly, the City of released 800,000 panoramic images as in 2017, allowing rapid processing of street features to aid urban development and maintenance planning. In navigation enhancements, companies like integrate Mapillary's crowdsourced data into their Map Creator platform, improving route accuracy and real-time updates for millions of users, while OsmAnd employs a Mapillary to provide street-level views for offline route planning. As of 2025, Mapillary's global volunteer community has contributed over 2 billion images across 190 countries, demonstrating significant scale in collaborative mapping. Key partnerships, including those with US for camera distribution programs and for disaster resilience initiatives, have amplified these efforts, with examples like BikeOttawa's capture of 450,000 images over 2,000 kilometers supporting local advocacy and planning. These collaborations underscore Mapillary's dual role in grassroots and market-driven applications, occasionally facilitated by tools like the Mapillary Tasker for coordinated image uploads.

Tools and Extensions

Mapillary Tasker

The Mapillary Tasker is a collaborative tool designed to coordinate efforts in enhancing street-level and by assigning specific tasks within defined geographic areas. Launched in beta on November 28, 2017, it enables users to create and join tasks focused on addressing coverage gaps through image capture, features, or verifying AI-detected objects. This tool facilitates targeted mapping improvements by breaking areas into manageable zones, allowing contributors worldwide to participate without needing prior coordination. Key features include task creation, where organizers define objectives and boundaries via the web app's navigation menu; progress tracking, visualized through color-coded grids that shift from red (incomplete) to green (covered) for capture tasks; and support for map editing that integrates with external platforms like (OSM) to add or refine features based on imagery. Verification tasks leverage the Verifier tool to confirm detections such as traffic signs, ensuring data accuracy before incorporation into broader mapping projects. In its initial beta phase, new task requests required admin approval to maintain quality and focus. Following Mapillary's acquisition by Meta Platforms in June 2020, the tool was listed under web and notifications features, but no further updates have been documented as of November 2025.

Mobile and Web Interfaces

Mapillary provides dedicated mobile applications for iOS and Android devices, enabling users to capture street-level imagery, upload sequences, and explore the platform's global dataset directly from smartphones. The iOS app, initially launched in 2014, has undergone iterative enhancements focused on usability and capture efficiency. In 2024, version 6.0.0 introduced a redesigned homepage, support for local sequences, leaderboards, and full light/dark mode compatibility, while version 6.1.0 added fixed-focus capture modes for higher speeds, manual capture options, lens toggling, and volume button integration for hands-free operation. By 2025, updates emphasized image handling and performance: version 6.6.0 in May accelerated uploads by 20% with a new uploader and added individual capture scheduling; version 6.8.0 in October introduced support for importing images captured by third-party camera apps or external devices, accessible via the home view, alongside minor UI refinements for smoother navigation. These enhancements streamline the workflow for contributors, particularly in dynamic environments like urban mapping. The app, redesigned in early for improved ease of use and faster capturing/uploading, continued evolving to match iOS capabilities. The version 6.4.32 update incorporated wide-angle camera support, manual capture outside sequences, flash toggling, and real-time distance displays, with version 6.4.84 adding infinity focus for driving scenarios and single-image deletion. In 2025, version 6.10.0 in October expanded the camera map menu for customizable map styles, centering, capture age visualization, and compass modes, while improving tap interactions and extending screen dimming for prolonged sessions. Both apps support background uploading, allowing captures to proceed offline before syncing when connectivity returns, which facilitates fieldwork in remote areas. Complementing the mobile apps, Mapillary's web interface serves as a central for broader interaction, accessible via any . Users can through integrated tools, though large-scale submissions are optimized via the companion uploader; the web app handles profile management, sequence oversight, and basic uploads for smaller sets. Exploration features enable seamless navigation of the global repository, with tools to zoom, filter by date or coverage, and view extracted map features like traffic signs or . Analytics are embedded through computer vision-derived insights, allowing users to assess coverage density, feature distributions, and contribution impacts via interactive maps and organizational . For organizations, the provides dedicated views to edit profiles, query contributions, and export data, enhancing collaborative oversight. Accessibility is prioritized across interfaces to broaden participation. Both mobile apps support multiple languages, with users able to adjust settings for interfaces in languages such as English, , Japanese, and , ensuring global usability. The web app follows standard guidelines, including keyboard navigation and compatibility for exploration views. Efforts to enhance overall user-friendliness, including simplified onboarding and reduced permissions prompts, have been ongoing since 2022. Offline capture in mobile apps, combined with queued uploads, supports intermittent , though full exploration requires an internet connection.

References

  1. [1]
    About Mapillary
    Mapillary brings together a global network of contributors who want to make the world accessible to everyone by visualizing the world and building better maps.
  2. [2]
    An Introduction to Mapillary
    Dec 11, 2024 · Mapillary is a crowdsourced street level imagery platform powered by computer vision. Anyone can upload imagery to Mapillary, using just about any camera or ...
  3. [3]
    Mapillary - Crunchbase Company Profile & Funding
    Legal Name Mapillary AB ; Also Known As Mapillary ; Operating Status Active ; Company Type For Profit ; Founders Jan Erik Solem, Johan Gyllenspetz, Peter Neubauer, ...
  4. [4]
    Mapillary Celebrates 10 Years and 2 Billion Images!
    Jan 12, 2024 · For a brief reflection on 10 years of Mapillary, we asked one of Mapillary's founders, Jan Erik Solem, to tell us about the experience of ...
  5. [5]
    Mapillary, the crowdsourced database of street-level imagery, has ...
    Jun 18, 2020 · Mapillary also says its “commitment to OpenStreetMap stays.” Writes Mapillary co-founder and CEO Jan Erik Solem (who also previously founded ...Missing: founders | Show results with:founders
  6. [6]
    https://techcrunch.com/2018/09/13/mapillary-rekognition-amazon/
  7. [7]
    Mapillary | FHWA - Department of Transportation
    Jun 1, 2021 · Mapillary is a street-level imagery platform. For transportation agencies, the tool offers a cloud-hosted solution for right-of-way imagery and pavement ...
  8. [8]
    First Photo Day - A Look at Two Years of Mapillary
    Oct 8, 2015 · When we started out in September 2013, there were three of us; me, Yubin and Johan. We started hacking pieces together in a co-working space in ...
  9. [9]
    Manifesto - Mapillary
    That's literally a long way from where we started in 2013. Here's the founding manifesto we wrote when we first started out. At Mapillary we want to create ...
  10. [10]
    Now Supporting Panoramas and Photo Spheres - The Mapillary Blog
    10 Sep 2014. Today we are adding support for uploading panoramas to Mapillary. With the advance of 360 degree (action) cameras and smartphone ...
  11. [11]
    Mapillary Raises $1.5M Seed Round
    Jan 15, 2015 · We just closed a $1.5M seed round led by Sequoia. Also participating in the seed round are some great funds, Playfair, Wellington, ...Missing: ABC | Show results with:ABC
  12. [12]
    Announcing Our $8M Series A Round - The Mapillary Blog
    Mar 3, 2016 · We just closed an $8 million Series A financing round to enable us to expand. Last year was an incredible year for our community and for ...Missing: ABC | Show results with:ABC
  13. [13]
    Scaling up with a $15M Investment - The Mapillary Blog
    Apr 12, 2018 · We're announcing an investment round that will let us push even further on developing top-ranking computer vision technology and building a ...Missing: B | Show results with:B
  14. [14]
    A Look Back at 2019 and the Road to One Billion Images
    Dec 16, 2019 · The Mapillary platform now hosts one billion high-resolution images from around the world, all open and available for improving maps.Missing: reaches August 2023
  15. [15]
    Mapillary Joins Facebook on the Journey of Improving Maps ...
    Jun 18, 2020 · We're very excited about the future and can't wait to build it with you. The Mapillary team. /Jan Erik, CEO and co-founder.Missing: founders | Show results with:founders<|control11|><|separator|>
  16. [16]
    Facebook acquires crowdsourced mapping company Mapillary
    Jun 19, 2020 · Mapillary Chief Executive Officer Jan Erik Solem, who founded the Malmö-based startup after leaving Apple Inc in 2013, said his company's ...Missing: September | Show results with:September
  17. [17]
    Facebook buys start-up in latest push to take on Apple and Google
    Jun 19, 2020 · Facebook has acquired Swedish mapping start-up Mapillary as part of an effort to take on Apple and Google at mapping the world at ground level.
  18. [18]
    Facebook Buys Swedish Mapping App - Nasdaq
    Jun 20, 2020 · Facebook (NASDAQ: FB) has inked a deal to acquire Mapillary, a Swedish company building a global street-level imagery platform.
  19. [19]
    Mapillary
    Mapillary is the platform that makes street-level images and map data available to scale and automate mapping.About Mapillary · Web explorer · How Mapillary works · Build with MapillaryMissing: official | Show results with:official
  20. [20]
    Mapillary release notes
    Mapillary release notes. 24 October 2025. Mapillary for iOS 6.8.0. Added support for importing images that have been captured using other camera apps or ...
  21. [21]
    How can I upload imagery to Mapillary?
    Dec 10, 2024 · The simplest way to contribute images is to use the Mapillary mobile apps which let you both capture and upload conveniently with your phone.
  22. [22]
    Mapillary Android app: the complete guide
    Dec 10, 2024 · The Android app lets you capture, upload, and explore images on Mapillary—all with your phone! This guide will get you set up with everything ...📸capturing · 📤uploading · 🗺️exploring The Map<|control11|><|separator|>
  23. [23]
    Mapillary iOS App: the complete guide
    Dec 10, 2024 · The iOS app lets you capture, upload, and explore images on Mapillary—all with your iPhone! This guide will get you set up with everything ...📸capturing · 📤uploading · 🗺️exploring The Map
  24. [24]
    The Complete How-To Guide on getting great Mapillary imagery
    Dec 10, 2024 · In this article, you'll find some extra guidelines as well as tips specific to capturing on foot, by bicycle, or by car.📸what And What Not To... · Do Not Capture · 🚘driving Tips
  25. [25]
    What is the best distance between uploaded photos? - Mapillary forum
    Apr 8, 2017 · For the purposes of great computer vision, it would be ideal to have about 5 m between the images.
  26. [26]
    Mapillary is powered by world-class computer vision
    Mapillary uses Structure from Motion to find the relative positions of images and create smooth transitions between them. Our OpenSfM pipeline is available as ...Missing: processing | Show results with:processing
  27. [27]
    OpenSfM
    Mapillary. OpenSfM. An open-source Structure from Motion library that lets you build 3D models from images. Get started with OpenSfM. Images provided by Vexcel.
  28. [28]
    Open Sourcing 3D City Reconstruction - The Mapillary Blog
    Nov 14, 2014 · The technique is called 'Structure from Motion' (SfM) and means that you compute the relative camera positions and a 3D reconstruction of the ...
  29. [29]
    Object detections - Mapillary
    Dec 10, 2024 · One of the key aspects of the Mapillary platform is the computer vision technology embedded into our imagery processing pipeline. Using a ...
  30. [30]
    How Mapillary is Ramping up Traffic Sign Recognition Globally
    Jun 15, 2018 · Our new traffic sign recognition algorithm supports 1,500 traffic sign classes—an increase of 600 classes—across over 100 countries, ...Missing: lanes | Show results with:lanes
  31. [31]
    Mapillary Vistas Dataset
    The Mapillary Vistas Dataset is a diverse street-level imagery dataset with 25,000 high-resolution images, 124 semantic object categories, and 100 instance- ...
  32. [32]
    [PDF] The Mapillary Vistas Dataset for Semantic Understanding of Street ...
    The Mapillary Vistas Dataset is a novel, large- scale street-level image dataset containing 25000 high- resolution images annotated into 66 object ...
  33. [33]
    Denser 3D Point Clouds in OpenSfM - The Mapillary Blog
    Oct 31, 2016 · At Mapillary we build and use OpenSfM to find the relative positions of images and create smooth transitions between them. That process is ...
  34. [34]
    Tools - Mapillary
    Mapillary apps and tools capture, upload, and use street-level imagery and data. Contribute imagery, manage map data, and bring it all to your mapping and GIS ...
  35. [35]
    Exploring point map features with the Mapillary web app
    Dec 10, 2024 · Currently, Mapillary generates point features for 42 object classes. You can explore and filter on these object classes via the Mapillary web ...Missing: explorer | Show results with:explorer
  36. [36]
    Images and Data for OpenStreetMap - Mapillary
    Mapillary maintains integrations with Rapid Editor, iD Editor, and JOSM for viewing imagery and AI extracted point data including traffic signs, street assets, ...Edit The Map With Mapillary · Tutorial: Using Mapillary In... · How Others Use Mapillary For...
  37. [37]
    Editing in Map Creator with Mapillary Imagery
    May 18, 2018 · Mapillary imagery can be viewed directly in Map Creator by clicking the Mapillary icon in the bottom right panel. Enabling the Mapillary layer ...Enabling Mapillary Imagery · Points Of Interest · Road Characteristics
  38. [38]
    ArcGIS - Mapillary
    Extract map data. Save time and money with automatic feature extraction that detects traffic signs and other street and road assets, and places them on the map.<|separator|>
  39. [39]
    Mapillary ArcGIS Pro add-in
    Mar 26, 2025 · The Mapillary add-in makes editing feature classes, updating attribution, and validating the existence of assets based on street-level imagery easy.✏️learning The Basics · 📍digitizing With Imagery · Selecting A Feature Class To...
  40. [40]
    500 million images available to everyone - The Mapillary Blog
    Apr 5, 2019 · The now 500 million images on Mapillary are public and available for everyone, all of them processed with computer vision to create map data at scale.
  41. [41]
    Open Street Level imagery for US with Mapillary – Said ... - YouTube
    Jan 30, 2025 · Mapillary is on the way to 3 billion images by 2025. Partnering with OSM US, we're expanding coverage across the US.Missing: milestone 2024
  42. [42]
    a dataset of crowdsourced street-level imagery annotated by road ...
    Sep 25, 2024 · As of January 2024, Mapillary contains about 170 million images in Germany, including hundreds of thousands for every major city, with over 50% ...
  43. [43]
    A Large Crowdsourced Street View Dataset for Mapping Road ...
    Jun 13, 2025 · Mapillary data: The Mapillary street view data has inherent limitations, including gaps in coverage, particularly in rural or remote areas, as ...
  44. [44]
    Harnessing Mapillary for Open Mapping and Disaster Resilience in ...
    In 2025, we are taking targeted steps to expand our collaboration with biking communities, particularly those in Metro Manila and key regional ...Missing: expansions | Show results with:expansions
  45. [45]
    How MapSwipe's New Mapillary Integration is Transforming Mapping
    May 25, 2025 · MapSwipe now uses Mapillary for street-level imagery, allowing volunteers to see details like road surfaces and waste, and validate what is on ...Missing: Meta acquisition
  46. [46]
    Introducing Street-Level Imagery Projects to MapSwipe4Web!
    MapSwipe4Web now uses street-level imagery from Mapillary, where users spot details like road conditions, for detailed mapping, available on the web app.Missing: integration expansions
  47. [47]
    MapSwipe and Mapillary integration improves open mapping
    May 31, 2025 · We're thrilled to see the new MapSwipe and Mapillary integration featured in Mapillary's blog, highlighting how street-level imagery is ...
  48. [48]
    Vermont Opens Up 5 Years of Road Imagery for Improving ...
    Mar 28, 2018 · The Vermont Transportation Agency is the first State Department of Transportation (DOT) in the U.S. to bring their photo logs to Mapillary.Missing: acquisitions Arizona
  49. [49]
    Mapillary fixes maps with computer vision - Geospatial World
    Sep 24, 2018 · ” The Department of Transportation in Arizona, for instance, recently uploaded 4.7 million images covering all of their highways. People and ...
  50. [50]
    [PDF] Leveraging Street Level Imagery for Urban Planning - IIASA PURE
    These include image datasets uploaded by Vermont Department of Transportation (5 million images collected over 5 years) and the Arizona Department of ...
  51. [51]
    Planning the City of Tepic with New Technologies and Citizen ...
    Feb 8, 2017 · With Mapillary our photo mapping initiatives have been focused in zones near the Historical Center of Tepic, in some neighborhoods where IMPLAN ...
  52. [52]
    Citation - Mapillary
    The Mapillary Vistas Dataset for Semantic Understanding of Street Scenes. G. Neuhold, T. Ollmann, S. Rota Bulò, and P. Kontschieder. The Mapillary Vistas ...
  53. [53]
    Releasing the World's Largest Street-level Imagery Dataset for ...
    May 3, 2017 · The Mapillary Vistas Dataset enables us to build the next-generation street-level image recognition platform that is needed for autonomously ...
  54. [54]
    https://blog.mapillary.com/product/2017/05/03/mapillary-vistas-dataset.html
  55. [55]
    [PDF] The Mapillary Traffic Sign Dataset for Detection and Classification ...
    In this paper, we will focus on detection and fine-grained classification of traffic signs on a new dataset. Traffic signs are key map features for navigation, ...
  56. [56]
    https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123680069.pdf
  57. [57]
    https://www.mapillary.com/dataset/places/
  58. [58]
    Mapillary Metropolis Dataset
    Mapillary Metropolis is a city-scale, multi-modality benchmarking dataset for researching and evaluating computer vision algorithms in the context of 3D ...Missing: growth history count 2014 2023
  59. [59]
    Mapillary CrowdDriven Dataset
    A benchmark for visual localization in outdoor scenes. The dataset contains pairs of image sequences with ground truth poses obtained through human annotation.
  60. [60]
    The Mapillary Vistas Dataset for Semantic Understanding of Street ...
    The Mapillary Vistas Dataset for Semantic Understanding of Street Scenes · Figures and Tables · Topics · 1,341 Citations · 67 References · Related Papers ...
  61. [61]
    Mapillary Vistas Image Data Collection - Kaggle
    It is a large, diverse dataset of street-level images labelled for semantic segmentation and object detection. Contains 25,000 high-resolution images ...
  62. [62]
    Mapillary Vistas Validation for Fine-Grained Traffic Signs - arXiv
    Aug 4, 2025 · This dataset and evaluation framework pave the way for more reliable, interpretable, and scalable perception systems. Code and data are ...
  63. [63]
    SAM for Road Object Segmentation: Promising but Challenging - NIH
    Jun 10, 2025 · The Mapillary Vistas dataset (https://www.mapillary.com/dataset/vistas accessed on 1 February 2025) is considered one of the largest street ...<|separator|>
  64. [64]
    The State of Mapillary: An Exploratory Analysis - ResearchGate
    Oct 16, 2025 · In this study, we attempted to get a deep insight into the Mapillary project through an exploratory analysis from the perspective of contributors.
  65. [65]
    https://www.researchgate.net/publication/338159120_The_State_of_Mapillary_An_Exploratory_Analysis
  66. [66]
    License Update: Now Creative Commons Share Alike - Mapillary Blog
    Apr 29, 2014 · Today we are making modifications to the license and are switching to a Creative Commons Attribution-ShareAlike 4.0 License (CC BY-SA).Missing: content | Show results with:content
  67. [67]
    CC-BY-SA license for open data - Mapillary
    May 12, 2025 · All images on Mapillary are shared under a CC-BY-SA license, which in short means that anyone can look at and distribute your images, and even modify them a ...Missing: 4.0 | Show results with:4.0
  68. [68]
    OpenStreetMap compatibility - Mapillary
    Dec 10, 2024 · We have integrations with two of the main tools used for editing OpenStreetMap, JOSM and iD Editor, to make it as easy as possible to enhance edits.
  69. [69]
    Mapillary Terms of Use
    Feb 15, 2024 · These Terms of Use (“Terms”) apply to your access to and use of the websites, mobile apps and other services, features, APIs, and functionalityMissing: 2020 | Show results with:2020
  70. [70]
    Expanding Access to Public Street-Level Imagery in the US
    Jul 8, 2024 · A growing number of cities and transportation departments are using Mapillary to create open street-level imagery datasets.Missing: partnerships | Show results with:partnerships
  71. [71]
    Mapillary Privacy Policy
    Mar 26, 2025 · Effective March 26, 2025. This Privacy Policy explains how Mapillary (“Mapillary,” “we,” or “us”) collects, uses, and shares information ...Collection Of Information · Sharing Of Information · United States Residents
  72. [72]
    How to Use Mapillary Data in OpenStreetMap
    May 23, 2019 · Mapillary data can be used via an organization download or Pic4Review. In OpenStreetMap, load the data as custom map data and use the photo ...
  73. [73]
    How to use Mapillary point data in OpenStreetMap iD
    Jan 16, 2020 · Mapillary has long been integrated into OpenStreetMap editors, and we are evaluating how to increase the value of both the imagery contributed<|separator|>
  74. [74]
    OSM US & Mapillary Launch CompleteTheMap US
    Sep 2, 2025 · This initiative rewards new and returning Mapillary contributors for their efforts collecting street-level imagery with exciting prizes.
  75. [75]
    Mapillary in QGIS3 - Contributing and equipment
    Jun 4, 2018 · It is now possible to add a Mapillary coverage basemap for any location and any coordinate system and filter it by user and shot time.
  76. [76]
    Case Studies - Mapillary
    HERE partners with Mapillary to support community mapping efforts to improve HERE maps. Mapillary imagery is available in the Map Creator editing tool for ...Missing: urban | Show results with:urban
  77. [77]
    Measuring Pedestrian-Level Street Greenery with Mapillary Imagery
    Mar 12, 2025 · Our research demonstrates that street greenery should be measured and designed from a pedestrian's perspective, not just from above or from cars.
  78. [78]
    Detroit: the most covered city on Mapillary
    Nov 18, 2024 · 99.8% of all public roads in Detroit have been mapped, most several times over, making Detroit the best mapped city on Mapillary!Missing: planning | Show results with:planning
  79. [79]
    The Mapillary Vistas Dataset for Semantic Understanding of Street ...
    The Mapillary Vistas Dataset has 25,000 high-resolution street images annotated into 66 object categories, with instance-specific labels for 37 classes. It is ...
  80. [80]
    "streetscape" package in R: A reproducible method for analyzing ...
    Nov 21, 2024 · "streetscape" package in R: A reproducible method for analyzing open-source street view datasets and facilitating research for urban analytics.
  81. [81]
    Benchmarking City-Scale 3D Map Making with Mapillary Metropolis ...
    This data set is designed with the goal of creating a completely novel and complex benchmarking paradigm for training and testing computer vision algorithms in ...
  82. [82]
    COCO + Mapillary | ICCV 2019
    The goal of the joint COCO and Mapillary Workshop is to study object recognition in the context of scene understanding.
  83. [83]
    Winning the ECCV Semantic Segmentation Challenge for ...
    Aug 29, 2018 · Mapillary just won the Semantic Segmentation Challenge for Autonomous Navigation in Unstructured Environments, co-located with this year's European Conference ...
  84. [84]
    Crowdmapping on Street-Level Imagery with MapSwipe - HeiGIT
    Mar 11, 2025 · The crowdmapping app MapSwipe now integrates also street-level imagery. This new feature enables even mapping efforts that require the highest level of detail.
  85. [85]
    Open Street Level imagery for US with Mapillary and OSM US
    Jun 20, 2025 · Mapillary helps the community enrich and validate OpenStreetMap data. We formed partnership with OSM US and distributed GoPro MAX cameras to ...Missing: MapSwipe integration
  86. [86]
    Introducing the Mapillary Tasker: Collaborate with Anyone on ...
    Nov 28, 2017 · The Mapillary Tasker is a tool for connecting the Mapillary community to work together on predefined tasks in specific areas, using street-level images.Missing: partner | Show results with:partner
  87. [87]
    2017 Year in Review - The Mapillary Blog
    Dec 29, 2017 · Verification tasks, as well as capture and map editing tasks, can be set up in the new Mapillary Tasker, which connects contributors with ...
  88. [88]
    Mapillary - Wikipedia
    Mapillary was launched in 2013 and acquired by Meta Platforms, Inc. in 2020. It offers street level imagery similar to Google Street View. Mapillary.
  89. [89]
    Meta's open mapping mission - Mapillary
    Dec 9, 2024 · Today, people and organizations all over the world have contributed over two billion images toward Mapillary's goal of helping people understand ...Missing: milestone 2023 2024
  90. [90]
    Web and notifications - Mapillary Community Forum
    Discuss the workings of the Mapillary platform on the web, email notifications, stats, and leaderboards. For bug and issue reports, please email ...
  91. [91]
    Mapillary iOS app camera improvements
    Apr 8, 2025 · We are releasing a big update to the Mapillary Camera in the Mapillary iOS app which focuses on improved camera performance and usability.Missing: Meta | Show results with:Meta
  92. [92]
    Introducing the all-new Mapillary Android App
    Feb 19, 2024 · The Mapillary Android app has undergone a full redesign to improve ease of use and make capturing and uploading easier than ever.Missing: mobile | Show results with:mobile
  93. [93]
    Mapillary Desktop Uploader: the complete guide
    May 22, 2025 · The Desktop Uploader has a straightforward UI, making uploading to Mapillary as easy as dragging a folder or file into the application. Both ...🗂️adding Files For... · 🎞️uploading Images · 📹uploading Videos
  94. [94]
    Exploring traffic signs with the Mapillary web app
    Dec 10, 2024 · Traffic signs are supported on Mapillary as map features extracted from image detections. Currently, Mapillary recognizes 1500 traffic...Missing: lanes | Show results with:lanes
  95. [95]
    Mapillary for Organizations: the complete guide
    Dec 10, 2024 · There are three main avenues to uploading imagery to Mapillary: our mobile apps, the Desktop Uploader, and Mapillary Tools for the command line.🏘️ Creating An... · 👩🏻‍🤝‍👨🏽adding... · 📬downloading...
  96. [96]
    App language. How to change - Mapillary Community Forum
    Nov 24, 2019 · Hi, I have changed the language setting to Japanese, then app is shown in Spanish. And when I set to Spanish, then app is shown in Chinese.Missing: offline multi- accessibility
  97. [97]
    2022 with Mapillary Android
    Jan 10, 2023 · Core updates and improvements made to the app. The team's efforts highlights to increase the app's accessibility and user-friendliness, as well ...Missing: offline mode multi-