code and data

The FABA-Instruct dataset is designed to enhance facial action unit (AU) and emotion recognition. It includes 19,877 images with detailed annotations and instructions. Unlike existing datasets, FABA-Instruct provides both categorical and descriptive annotations, making it a valuable resource for training advanced AI models.

Car Crash Dataset (CCD) is collected for traffic accident analysis. It contains real traffic accident videos captured by dashcam mounted on driving vehicles, which is critical to developing safety-guaranteed self-driving systems. CCD is distinguished from existing datasets for diversified accident annotations, including environmental attributes (day/night, snowy/rainy/good weather conditions), whether ego-vehicles involved, accident participants, and accident reason descriptions. An overview of our accident annotations is dipicted in the figure.
RIT-18 dataset [Download]

RIT-18 is a novel compositional activity dataset collected by ACTION Lab at RIT containing 18 compositional activity classes. We collected video clips from 51 volleyball games on YouTube. With comprehensive annotations, RIT-18 is a large scale dataset for group activity understanding tasks such as group activity recognition, future activity anticipation, and rally-level winner prediction. The benchmark for these three tasks is provided.
BIT-Interaction dataset [Download]

BIT-Interaction dataset contains 400 videos of human interactions, distributed in 8 interaction classes, including handshake, kick, etc.