Giters

DefTruth / torchlm

💎A high level pipeline for face landmarks detection, it supports training, evaluating, exporting, inference(Python/C++) and 100+ data augmentations, can easily install via pip.

Home Page:https://github.com/DefTruth/torchlm

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

is PIPNet the SOTA ?

IQ17 opened this issue · comments

commented

Hi, thanks for the work!
I am from the industrial, and I plan to choose one algorithm and train with custom datasets.

I would like to ask which algorithm is the SOTA choice for the 2D facial landmarks task?
By SOTA I mean the video stability and accuracy given in the paper or github.

Also, any suggestions for 3D facial landmarks?

commented

torchlm offer a PyTorch re-implementation for PIPNet, this algorithm is SOTA in some common use datasets with CPU realtime FPS, you can check their paper for more details. For 3D facial landmarks, may be you can try 3DDFAV2、FaceMesh(from mediapipe), FAN etc.

commented

Thanks!

commented

From benchmark table, it is very similar to paper report. Have you train your PIPnet with augmentation? I think the performance should better

commented

From benchmark table, it is very similar to paper report. Have you train your PIPnet with augmentation? I think the performance should better

hi~, torchlm is not aims to build it's own SOTA algorithm but reproduce some SOTA models report by papers and other open source codes, so, i have re-implentment PIPNet carefully and make sure the users can use the SOTA's pretrained weights directly. I like it's Heatmap+Regression designs. The performance of PIPNet is source from their paper. It's very easy to fine tune PIPNet with the high-level APIs in torchlm, some transforms, such as MixUp, may boost the performance. Also, the inference codes with Pytorch/ONNXRuntime(Python), ONNXRuntime/MNN/NCNN/TNN(C++) have already release at torchlm. For the commercial reason, the models trained with my custom datasets will not be release.