oulutan / drone_fasterrcnn Goto Github PK

🐛 Bug

Dependencies in the Dockerfile are out of date. Realised after a heavy debugging session:

the source of miniconda.sh is out of date, used https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh instead.

Moreover, torchvision shall be installed via conda! (Also produces errors)

I recommend to revise all dependencies.

To Reproduce

Try to build the docker image from the file :)

Expected behavior

error

Environment

I ran on Ubuntu, but this ain't the point now.

❓ Questions and Help

Hi I am currently using ur pretrained model and inferencing on visdrone test dev dataset.Also same model I have incorporated into
online tracking application.I am not sure how to improve the performance for this model?could you guide?what are the instances I need to use?

❓ Questions and Help

I am a novice of CV, Can you provide your source code of video segmentation and combination? Like how to detect from video or you can give me some references to look at.

❓ Questions and Help

I was wondering how well your detector performs on the test set. Can you tell us your AP-score?

❓ Questions and Help