Commit graph

5 commits

Author SHA1 Message Date
KanagawaYamada
391c8797db
Fix Numpy 2.0, OpenCV 5.0, and ONNXRuntime GPU compatibility
1. Fix shape broadcasting crash in Numpy 2.0 with .flatten()

2. Fix cv2.solvePnP in OpenCV 5.0 by setting useExtrinsicGuess=False on the first frame

3. Switch default models to _gpu.onnx to avoid FusedConv Clip crashes on CUDAExecutionProvider
2026-08-29 19:34:59 +07:00
yamichan420
0204d8a4a7
Add files via upload
apparently, onnxruntime 1.9 and later require that the available providers be explicitly listed as an argument in InferenceSession()
or i was getting that error anyway, and it was preventing OpenSeeFace from working
i'm running Arch Linux using the build of python-onnxruntime available in the AUR
anyway, i edited tracker.py and retinaface.py to find the list of available providers and pass that argument
and that makes it work for me!
yay!
2022-03-29 22:21:40 -07:00
Emiliana
b0a61b41d5 RetinaFace inference is now done at full resolution without padding. 2020-03-22 18:01:21 +01:00
Emiliana
033a48b132 This changes the points used for PnP solving and replaces the RetinaFace model with a higher resolution version. 2020-02-28 02:10:23 +01:00
Emiliana
cd9cfd9ee8 RetinaFace was added as a face detection scheme. 2020-02-26 00:54:32 +01:00