facebookresearch/supervision-by-registration
Supervision-by-Registration: An Unsupervised Approach to Improve the Precision of Facial Landmark Detectors
PythonNOASSERTION
Issues
- 0
- 5
About how get the landmark of Reg
#57 opened by xyzhai-pro - 0
Where is the snapshot
#79 opened by dolphins22365 - 6
replace the backbone
#58 opened by AaronXinWu - 0
300VW Training
#78 opened by yongyongdown - 5
Face Mesh Results?
#72 opened by CodingMice - 0
Training on another landmark problems
#77 opened by eanemo - 11
Train on my viedo dataset
#62 opened by HansonXia - 3
How to debug LK Module
#74 opened by sljlp - 16
About the optimization of LK
#49 opened by zhang-zx - 2
Where pam is applied
#75 opened by makalo - 0
Torch Script
#76 opened by eusebioaguilera - 0
- 2
Where to get the pretrained model?
#64 opened by fanyhchn - 1
Pretrained Model
#71 opened by ShakedDovrat - 0
test on own video
#70 opened by fnhdx - 9
Result for pretrained model is very bad and unstable at angled face. Is the pretrained model that bad?
#69 opened by yxchng - 0
About using camera
#68 opened by come880412 - 4
Training problem in the first stage
#66 opened by KeyKy - 1
Problem about training on a small trainset
#67 opened by JINXvvv - 1
Base Detector?
#63 opened by ModarD - 6
What is the size of the training model?
#61 opened by hyb1234hi - 2
maybe something wrong in transforms.py
#53 opened by momiji7 - 8
Train my own dataset
#55 opened by abdou31 - 1
the LK operation is very slow, is this normal?
#54 opened by taozhiqi - 1
- 5
confused about the NMS
#45 opened by wqz960 - 2
what's the meaning of meta files?
#65 opened by JINXvvv - 3
The effect of landmark detection dramatically decreases as the distance increases
#59 opened by xfj81525 - 0
- 0
- 4
evaluate a single image
#51 opened by JiHaoqina - 0
To improve efficiency, you can change the backbone of our SBR, which is VGG-16 (see https://github.com/facebookresearch/supervision-by-registration/blob/master/lib/models/cpm_vgg16.py). Besides, we use an input size of 224, you can downsample the input to 128, or 96, or 64.
#50 opened by xfj81525 - 12
About face bounding box
#47 opened by pokidyshev - 4
Doubts about Video Tags
#48 opened by zrqi - 1
About speeding up the LK
#46 opened by zhang-zx - 1
evaluate the 300W video
#44 opened by cploveflower - 3
Can I use my own video to test the cpm-sbr?
#40 opened by dyther - 1
How to get the coordinates from the heatmaps?
#41 opened by SunnyDp - 5
How to make the model smaller?
#43 opened by zhang-zx - 6
Problems when training on 300W
#42 opened by zhang-zx - 4
- 1
can this project be transferred to mobile services
#38 opened by imerse - 1
The loss between 'batch_location' and the groundtruth
#37 opened by SunnyDp - 1
- 12
landmarks still shake after 10 epochs
#34 opened by Racha1992 - 1
Some questions about the different configs in the "specify_parameter" function in the cpm_vgg16.py
#35 opened by SunnyDp - 1
- 3
Error for eval.py
#31 opened by NikolayTV - 1
Some questions about preparing data
#32 opened by tongxinjie