search

experiencor / keras-yolo3

Training and Detecting Objects with YOLO3
MIT License
1.6k stars 861 forks source link

Fix requirements and Dockerize #311

Closed maininformer closed 3 years ago

maininformer commented 3 years ago

Hello, Thanks for this implementation. I was going to use this but came across some roadblocks:

Using Python, at least, 3.9, pip install for these requirements will fail; tensorflow 1.15.0 for instance is not supported on Python 3.9 yet because it's dependencies are not fully supported yet, see here.

Moreover it is best to leave the dependency resolution to the main packages themselves (Keras for instance will install numpy and scipy for us).

I made added a Dockerfile with Python 3.7 that will successfully install all the dependencies.(pip freeze here will give the same output as the original requirements.txt.

Also added instructions for a docker run.

maininformer commented 3 years ago

closing due to found compatiblity issue between <Keras-Applications==1.0.8 and tensorflow==1.15.0