The official implementation of our ICCV 2023 paper "Objects do not disappear: Video object detection by single-frame object location anticipation.

License: MIT License

Makefile 0.10% C++ 6.71% C 5.40% Python 81.47% Cython 5.89% Shell 0.42%

video-object-detection-by-location-anticipation's People

Contributors

Stargazers

Watchers

Forkers

geekiter

video-object-detection-by-location-anticipation's Issues

Backbone

Your work is very valuable. I would like to ask you some questions about Backbone. Where can I obtain the code for your DETR architecture and Resnet101? I see that the code presents the training process for Resnet50.

What is the resolution of the input frames at the testing and training stages?

training problem

Thank you very much for providing the code for your paper.
However, I noticed that I am unable to generate any weight files during training.
Additionally, if possible, could you provide pre-trained weight files that can be directly used to validate accuracy？

How to prepare the training data?

Hi! Thank you very much for providing the code for your paper. However, there seems to be no detailed description of data processing in the code base you provided, only the data path is marked in the code. Therefore, I would like to ask you about the specific data processing steps. Can you help me?

Recommend Projects