Event-based Vision Datasets

This page contains our realeased event-based datasets. Please star this repository if you use our datasets.

1. $\text{THU}^\text{HSEVI}$ Dataset

• Our $\text{THU}^\text{HSEVI}$ dataset could be from https://drive.google.com/file/d/1im7Idfx7VP8c1krIdyS7zeiKTmgsU6qR/view?usp=sharing or from https://pan.baidu.com/s/13GVxDUc_81_DedRnIHIFSA (extract code: 69g5).
• The dataset is corresponding to the IEEE T-PAMI 2022 paper "SuperFast: 200× Video Frame Interpolation via Event Camera". Code is available at: https://github.com/lisiqi19971013/SuperFast.

2. E-3DTrack Dataset

• Our E-3DTrack dataset is available now at: https://drive.google.com/file/d/1gOlNqq7FaRbA9MK8YYUBozxf9a3cYG-S/view?usp=sharing or https://pan.baidu.com/s/1htePOwKzifQURK1gCWu3qQ (extract code: 2024).
• The dataset is corresponding to the CVPR 2024 paper "3D Feature Tracking via Event Camera". Code is available at: https://github.com/lisiqi19971013/E-3DTrack.

3. $\text{THU}^\text{E-OccVP}$ Dataset

• Our $\text{THU}^\text{E-OccVP}$ dataset is available now at: xxx.
• The dataset is corresponding to the IJCV 2025 paper "RGB-D Visual Perception for Occluded Scenes via Event Cameraa". Code is available at: xxx.

4. $\text{THU}^\text{E-HRSAI}$ Dataset

• Our $\text{THU}^\text{E-HRSAI}$ dataset is available now at: https://cloud.tsinghua.edu.cn/d/9257d56a708d46b7a053/. This dataset is compressed into four parts, totaling 34.4 GB, and requires approximately 467 GB of disk space for extraction.
• The dataset is corresponding to the Information Fusion 2025 paper "High-Resolution Synthetic Aperture Imaging Method and Benchmark based on Event-Frame Fusion". Code is available at: https://github.com/lisiqi19971013/E-HRSAI.

5. $\text{THU}^\text{ERGB-SAI}$ Dataset

• Our $\text{THU}^\text{ERGB-SAI}$ dataset is available now at: https://cloud.tsinghua.edu.cn/d/a7c74df9c1814779b705/. This dataset is compressed into four parts, totaling 35.8 GB, and requires approximately 508 GB of disk space for extraction.
• The dataset is corresponding to the SCIENCE CHINA Information Sciences 2025 paper "Event-Enhanced Synthetic Aperture Imaging". Code is available at: https://github.com/lisiqi19971013/THU-ERGB-SAI.

6. Occlusion-400 Dataset

• Our Occlusion-400 dataset is available now at: https://pan.baidu.com/s/1o2nTtNFgeA-Feri1OaGETw (extract code: ddqf) or https://cloud.tsinghua.edu.cn/f/ece1ded9ade04a7291a2/?dl=1.
• The dataset is corresponding to the MIR 2022 paper "Image De-Occlusion via Event-Enhanced Multi-Modal Fusion Hybrid Network". Code is available at: https://github.com/lisiqi19971013/Event_Enhanced_DeOcc.

7. EventSR Datasets

• The event stream super-resolution datasets are available at:
  • N-MNIST: https://drive.google.com/file/d/19VNS5gJBHyKKCzsyg9OlBrw3o1dx5UM2/view?usp=sharing
  • Cifar10-DVS: https://drive.google.com/file/d/1od7m1AUA6YinG7qbcXU0pNWWYFSlP37g/view?usp=sharing
  • ASL-DVS: https://drive.google.com/file/d/17E7Doq3F9Cn-QdGlJLmEvHlN5KMs-h9f/view?usp=sharing
  • Event Camera Dataset: https://drive.google.com/file/d/1IlaacDk56pNVHLLHJWFrghZvruaLP-iK/view?usp=sharing
• These datasets are corresponding to the ICCV 2021 paper "Event Stream Super-Resolution via Spatiotemporal Constraint Learning". Code is available at: https://github.com/lisiqi19971013/EventStream-SR.