A Biologically Inspired Separable Learning Vision Model for Real-time Traffic Object Perception in Dark

简体中文 | English

Dark-traffic benchmark

(object detection, instance segmentation, and optical flow estimation in the low-light traffic conditions)

images (~10k) and annotations (~100k) in Google Drive

The images are now available, and annotations will be released after the paper is processed by the journal/conference.

• 9.05 update: Available on Expert Systems with Applications, annotations (Detection/Segmentation/Flow) is released.
• 9.07 update: Code (SLVM for Static & Motion perception) is preparing to be released.
• 9.09 update: Code (SLVM for Motion perception) is released.
• 9.24 update: Code (SLVM for Static perception) is released.

SLVM for Det&Seg

training step coming soon...

SLVM for optical flow

Re-produce

    python pip install -requirements.txt

References

• https://github.com/AlanLi1997/slim-neck-by-gsconv
• https://github.com/AlanLi1997/rethinking-fpn
• https://github.com/ultralytics/ultralytics
• https://github.com/haofeixu/gmflow
• https://github.com/neufieldrobotics/NeuFlow_v2