error at startup

[ghost]

###################### Now Running the X 1 task ##############################

############################################################################### ####################### Step 1: Crop and Align Face ########################### ###############################################################################

Crop and Align 0000000043.png image Traceback (most recent call last): File "/content/dfdnet//test_FaceDict.py", line 187, in align_and_save(ImgPath, SavePath, SaveInput, SaveParam, UpScaleWhole) File "/content/dfdnet//test_FaceDict.py", line 46, in align_and_save source = get_5_points(img) ^^^^^^^^^^^^^^^^^ File "/content/dfdnet//test_FaceDict.py", line 26, in get_5_points dets = detector(img, 1) ^^^^^^^^^^^^^^^^ RuntimeError: Error while calling cudaOccupancyMaxPotentialBlockSize(&num_blocks,&num_threads,K) in file /root/.cache/uv/sdists-v7/pypi/dlib/19.24.2/hPJjJZ9cPqqfbTFto82d9/src/dlib/cuda/cuda_utils.h:164. code: 222, reason: the provided PTX was compiled with an unsupported toolchain.

I can’t fix it in any way, can someone tell me what to do about it? This error appears when working with T4 GPU Google Colab

[wearkxx]

Reinstall python version from 3.12 to 3.10, and then install dlib-binary instead of regular dlib, so that when you run the DFDNet, the images are cropped and faces are detected on them through the CPU, not through the GPU. After that, install an older version of NumPy to avoid errors.