Train recognition models for street scenes
Develop your algorithms with diverse data from all over the world, including different weather, season, time of day, camera, and viewpoint conditions.
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 of control points, depicting urban and suburban scenes with challenging appearance and viewpoint changes
Learn moreMapillary Metropolis Dataset
A multitask benchmarking framework comprising complementary data modalities at a city-scale size, registered across different representations, and enriched with human and machine generated annotations.
Learn moreMapillary Planet-Scale Depth Dataset
A diverse dataset of 750'000 street-level images with metric depth information for outdoor metric depth estimation.
Learn moreMapillary Street-level Sequences Dataset
A benchmark dataset for lifelong place recognition from image sequences. 1.6 million images from diverse geographies and scene characteristics, provided with GPS coordinates and sequence information.
Learn moreMapillary Traffic Sign Dataset
A benchmark dataset with bounding box annotations for detecting and classifying traffic signs around the world. 100,000 images with over 300 traffic sign classes, with manual and machine annotations
Learn moreMapillary Vistas Dataset
A benchmark dataset of manually annotated training data for semantic segmentation of street scenes. 25,000 images pixel-accurately labeled into 152 object categories, 100 of those instance-specific.
Learn moreStart accessing imagery and map data
Mapillary is the platform that makes street-level images and map data available to scale and automate mapping. We’re committed to building a global service for everyone.