For the compatibility with the latest transformers lib.

Fix the bug by latest transformers and update req.txt.

Add half precision setting in README.

Align lib_name as birefnet and add inference endpoint option.

Dtype adaptability between FP32 and FP16 in inference.

Update the bibtex.

Update birefnet.py

Update README.md

Update config.json

Rename BiRefNet_T to BiRefNet_lite.

Correct the `_name_or_path ` in config.json.

Correct `_name_or_path` in config.json.

Initial uploading of the BiRefNet for general use with swin_v1_tiny as the backbone.

initial commit