SthPhoenix

> Hey @jinzaz, not sure why you tagged me. I am not affiliated with the project. The error says something about an invalid argument given to a C/C++ function through...

> @SthPhoenix i use the deploy_cpu.sh , Because my region can't access drive.google.com , So I download the corresponding onnx file in the directory in advance, and then execute deploy_...

Just run for example `md5sum glintr100.onnx` on downloaded models. It should be: ``` 50febd32caa699ef7a47cf7422c56bbd scrfd_2.5g_gnkps.onnx 3b366b98f786426f79629ddb2e56629c glintr100.onnx ``` If it's different - you need to re-download the models

Hi! I haven't tested RTSP streams, and haven't used RTX2080 for a while. On RTX3090 API can handle up to around 350-450 fps, though I would recommend to use processing...

Hi! Main goal of this repo is face recognition. Those models make no improvements in terms of recognition - recognition module uses 5 key points for recognition, so 106 and...

Yes, as I said before, it can be done, but will break performance. Though I'll keep in mind that this feature might be useful.

Hi! You can use snippet I have provided [here](https://github.com/SthPhoenix/InsightFace-REST/issues/101#issuecomment-1309138006) as reference Though I should mention that code you provided refers to original repo, which is quite different from code provided...

Hi! I wasn't able to reproduce this bug with np.float depreciation, I have rebuild image without cache, but can't see this bug. Can you try building again from this repo...

Hi! That's weird, I have build image from scratch and have no errors, but anyway I'll bump versions, just in case.Thanks!

Hi! What `max_size` and `det_model` arguments are currently used? This issue mostly influence original `scrfd_*g_bnkps` models at higher resolution (640x640 and above), to partially solve this issue `scrfd_*g_gnkps` models were...