Shu Kong, Ph.D. | Visual Intelligence Lab

We have released our benchmarking protocols and datasets to support further research in the community. We hope they prove valuable to a wide audience. Please follow the links below for more details.

	AutoExpert protocol for data annotation based on expert-crafted guidelines for autonomous driving
	Data retrieved from LAION-400M to study Retrieval-based Augmented Learning (RAL)
	A High-Resolution for Instance Detection with Multi-View Instance Capture
	Dataset to Study Instance Tracking in 3D Scenes from Egocentric Videos
	A Benchmark to Study Few-Shot Detection from Annotation Guidelines
	Data to Study Unstructured Unlabeled Optical Motion Capture
	AccessDB: A Dataset for Auto-detecting Inaccessibility of Everyday Objects
	A Benchmark to Study Long-Tailed 3D Detection
	UAL-Bench: A Benchmark for Unusual Activity Localization in Videos
	Dataset for Shoeprint Identification in Crime Scenes
	Dataset for Open-Vocabulary Part Segmentation
	Dataset for Image Aesthetics Rating
	Dataset for Fossil Pollen Recognition
	Dataset for C. elegans segmentation, body shape estimation, and age estimation
	Dataset for Pollen Grain Detection