timm==0.4.12 transformers fairscale==0.4.4 pycocoevalcap torch torchvision Pillow scipy git+https://github.com/openai/CLIP.git git+https://github.com/IDEA-Research/Grounded-Segment-Anything.git/#subdirectory=segment_anything git+https://github.com/xinyu1205/recognize-anything.git addict diffusers[torch] gradio huggingface_hub matplotlib numpy onnxruntime opencv_python pycocotools PyYAML requests setuptools supervision termcolor yapf nltk