Dataset Hub

The multimodal distracted driving detection dataset (2016‑2018) and the earlier dataset (2014‑2016) include detailed information such as the number of drivers and average driving time.

This dataset contains 4,983 weather images covering most severe weather conditions, divided into four categories: fog, rain, snow, and clear. Each category is split into a training set (4,000 images) and a test set (983 images). Images were primarily collected from the Internet and filtered according to specific requirements. Image sources include aerial photography, handheld cameras, news footage, traffic accident recordings, and dash‑cam recordings. The dataset is intended for recognizing weather conditions on traffic roads, so most images feature complex road scenes such as urban streets, highways, and rural roads.

Woven Traffic Safety (WTS) is provided by Woven by Toyota, Inc. to highlight detailed vehicle and pedestrian behaviors in various simulated traffic events, including accidents. The dataset contains over 1.2 k video events across more than 130 traffic scenarios, combining vehicle‑ego and fixed‑camera views. Each event includes a comprehensive textual description of observed behaviors and context. Additionally, about 4.8 k pedestrian‑related traffic videos from BDD100K are supplied with detailed textual annotations for external training and testing.