Deep-Live-Cam
Deep-Live-Cam copied to clipboard
hacksider
The program only uses the CPU for calculations on mac computers
I'm facing the same issue on my m3 max device too. I have following the steps and successfully run with python run.py --execution-provider coreml. But the app only use my cpu and thus very laggy
+1, is there a way to get better support on M3 chip? It's very laggy on my M3 macbook and I believe the GPU is not fully used.
Same question i came here to ask, its attrocious on mac, like severe lag, also it always uses continuity cam instead of the builtin mac cam, don't see a flag to disable it.
I'm facing the same issue on my m3 max device too. I have following the steps and successfully run with
python run.py --execution-provider coreml. But the app only use my cpu and thus very laggy
Here is the monitor data
Same question i came here to ask, its attrocious on mac, like severe lag, also it always uses continuity cam instead of the builtin mac cam, don't see a flag to disable it.
I changed the default camera by editing the ui.py file - look for camera = cv2.VideoCapture(0) and change the number. 0 was my iPhone, 1 was my monitor, I assume 2 will be my macbook built in one. Your numeration will vary.
I installed the above libraries.
How did you manage to get the onnxruntime-silicon package installed ?! I just get the error:
ERROR: Could not find a version that satisfies the requirement onnxruntime-silicon (from versions: none)
ERROR: No matching distribution found for onnxruntime-silicon
How did you manage to get the onnxruntime-silicon package installed ?! I just get the error:
ERROR: Could not find a version that satisfies the requirement onnxruntime-silicon (from versions: none) ERROR: No matching distribution found for onnxruntime-silicon
wrong python version?
Using a 32-bit ONNX model increases the inference speed to 15fps, primarily because onnxruntime doesn't yet fully support Apple's M-series chips.
- You can refer to this repository: Deep-Live-Cam.
- I am currently converting an ONNX model to a PyTorch model and utilizing MPS for acceleration, but I have encountered some issues during inference and am working to resolve them.
If you are interested, feel free to contact me via email: [email protected].
How did you manage to get the onnxruntime-silicon package installed ?! I just get the error:
ERROR: Could not find a version that satisfies the requirement onnxruntime-silicon (from versions: none) ERROR: No matching distribution found for onnxruntime-siliconwrong python version?
yeah it's newer than 3.10 I tried to downgrade python but didn't seem to "work" for me. I'm new to the world of Python though...
It looks like the package has been renamed to onnxruntime-arm64 now?
How did you manage to get the onnxruntime-silicon package installed ?! I just get the error:
ERROR: Could not find a version that satisfies the requirement onnxruntime-silicon (from versions: none) ERROR: No matching distribution found for onnxruntime-silicon
I installed miniconda, then used miniconda to create a python virtual environment, installed python3.10.10 in the virtual environment, and then executed the commands according to the author's instructions.
Perhaps a silly question... but are you on an Arm based mac or an older intel mac?
Perhaps a silly question... but are you on an Arm based mac or an older intel mac?
yeah, M2 Max here.
Using a 32-bit ONNX model increases the inference speed to 15fps, primarily because onnxruntime doesn't yet fully support Apple's M-series chips.
- You can refer to this repository: Deep-Live-Cam.
- I am currently converting an ONNX model to a PyTorch model and utilizing MPS for acceleration, but I have encountered some issues during inference and am working to resolve them.
If you are interested, feel free to contact me via email: [email protected].
Did you ever successfully finish your conversion @solstice-gao ?
