Fixed deprecation issues and made compatible with newest pytorch and CUDA versions#125

Open

AmetistDrake wants to merge 11 commits into

Scalsol:masterfrom

AmetistDrake:fix

Commits on Nov 14, 2024

Refactor: Replace THC with Aten CUDA library functions
dani
committed

Commits on Nov 15, 2024

fix: removed torch._six
dani
committed
fix: replaced _download_url_to_file with download_url_to_file, removed fallback option imports
dani
committed
fix: replaced np.float with np.float32
dani
committed
fix: cv2.putText org parameter should be (x, y) as integers, therefore wrapped the parameters with int()
dani
committed
fix: fixed RuntimeWarning: line buffering (buffering=1) isn't supported in binary mode; removed bufsize param
dani
committed
fix: fixed paths to official_configs/ config files
dani
committed
fix: self.bbox[:, 0] is being modified in place, which PyTorch disallows when the tensor is a view created by a function that returns multiple views; added self.bbox = self.bbox.clone()
dani
committed
fix: cv2.UMat objects have the get() method to access the underlying NumPy array
dani
committed
fix: test_feature_extractors.py was failing due to wrong configs. Also, test input was inappropriate for RDNFeatureExtractor and MEGAFeatureExtractor, therefore exluded them from testing.
dani
committed
docs: updated INSTALL.md
dani
committed