#AutonomousDriving #DashCamVideo #Datasets #RoadScenarios #maadaa.ai
With the increasing homogenization of autonomous driving sensor solutions and computing platforms, “scenario-based data” driven algorithm optimization is becoming the road to commercializing autonomous driving technology.
From the perspective of autonomous driving algorithms, the larger the number of training datasets, the wider data collection, the more accurate data annotation, the more significant data classification, the more diverse types, the better the performance of the trained autonomous driving algorithms.
In other words, without the support of sufficient quantity and high-quality data, autonomous driving lacks quantitative perception of the vehicle’s surroundings, just like a person without eyes, the decision-making system of autonomous driving cannot work properly. Not to mention, compared to other application scenarios, autonomous driving needs to face more complex and changing road conditions.
maadaa.ai is providing a series of dash cam video datasets, covering variable road scenarios in several Asia cities, under different weather conditions.
Version 1.0 dataset provides videos of a total length of around 6,300 minutes, with more than 20k annotated images, including 200k bounding boxes and attributes such as occlusion, orientation, etc.
- Variable road scenarios, including highways, crossroads and other road scenes.
- A variety of weather conditions, including sunny days, rainy days and others.
- Comprehensive labeling of objects, including cars, humans, electric bicycles, buses, trucks and other 10 labeling categories, including most types of road.
1. Sunny Day Crossroads Dash Cam Video Dataset（MD-Auto-008）
MD-Auto-008 datasets have about 10k annotated images. The video length is about 480minutes. Driving Recorders Images. The resolution is over 1920 x 1080 and the number of frames per second of the video is over 34. Total more than 10 typical object categories, such as human, car, electric bicycle, van, truck, etc.
Link: https://maadaa.ai/dataset/sunny-day-crossroads-dash-cam-video-dataset/
2. Sunny Day City Road Dash Cam Video Dataset (MD-Auto-007)
MD-Auto-007 datasets have about 4.5k annotated images, Video length is about 300minutes. Driving Recorders Images. The resolution is over 1920 x 1080 and the number of frames per second of the video is over 33. Total more than 10 typical object categories, such as human, car, electric bicycle, van, truck, etc.
Link:https://maadaa.ai/dataset/sunny-day-city-road-dash-cam-video-dataset/
3. Cloudy Day Crossroad Dash Cam Video Dataset (MD-Auto-010)
MD-Auto-010 datasets have about 2.4k annotated images. The video length is about 120minutes. Driving Recorders Images. The resolution is over 1920 x 1080 and the number of frames per second of the video is over 32.Total more than 10 typical object categories, such as human, car, electric bicycle, van, truck, etc.
Link: https://maadaa.ai/dataset/cloudy-day-crossroad-dash-cam-video-dataset/
4. Cloudy Day City Road Dash Cam Video Dataset (MD-Auto-009)
MD-Auto-009 has about 1k annotated images. The video length is about 60minutes. Driving Recorders Images. The resolution is over 1920 x 1080 and the number of frames per second of the video is over 31. Total more than 10 typical object categories, such as human, car, electric bicycle, van, truck, etc.
Link: https://maadaa.ai/dataset/cloudy-day-city-road-dash-cam-video-dataset/
5. Low lighting Dash Cam Video Dataset (MD-Auto-011)
MD-Auto-011 has about 800 annotated images. The video length is about 60minutes. The resolution is over 1920 x 1080 and the video’s number of frames per second is over 30. Mainly from Low lighting, crossroad, avenues and paths as the main scene. The labels include human, car, electric bicycle, van, truck, etc.
Link: https://maadaa.ai/dataset/low-lighting-dash-cam-video-dataset/
6. Rainy Dash Cam Video Dataset (MD-Auto-012)
MD-Auto-012 has about 6.4k annotated images. The data collection is driving recorder images. The resolution is over 1920 x 1080 and the number of frames per second of the video is over 30. The annotation types are bounding boxes and tags. The application scenario is autonomous driving.
Link: https://maadaa.ai/datasets/DatasetsDetail/Rainy-Dash-Cam-Video-Dataset
Further Reading:
AI for virtual fitting: inspired by datasets （Open & Commercial）
AI for fake Detection in Fashion and E-commerce industries: The related open & commercial datasets
Face Parsing: use cases and open datasets
