search

JACKYLUO1991 / FaceParsing

EHANet: An effective hierarchical aggregation network for face parsing
MIT License
169 stars 31 forks source link

How to test on image #13

Closed myounus96 closed 4 years ago

JACKYLUO1991 commented 4 years ago

@myounus96 You should first configure the relevant parameters in parameters.py, and then run the main.py script. Of course, first you need to train the model...

myounus96 commented 4 years ago

@myounus96 You should first configure the relevant parameters in parameters.py, and then run the main.py script. Of course, first you need to train the model...

Have done that..tester.py only producing accuracy metrics..not coloured masked predicted images..

JACKYLUO1991 commented 4 years ago

@myounus96 Please refer to the detection.py script in the deployment folder.

myounus96 commented 4 years ago

it just keep giving 'load NONE from pretrained checkpoint' error.

JACKYLUO1991 commented 4 years ago

@myounus96

You should load face detect box weight from （https://pan.baidu.com/s/14MzwqOBxY2tciC5V8MXW2w and password is a9ty）first.

myounus96 commented 4 years ago

@myounus96

You should load face detect box weight from （https://pan.baidu.com/s/14MzwqOBxY2tciC5V8MXW2w and password is a9ty）first.

Thank you will try this.

myounus96 commented 4 years ago

I got 'page doesn't exist'. image

JACKYLUO1991 commented 4 years ago

@myounus96 Please check your email.

myounus96 commented 4 years ago

@JACKYLUO1991 didn't receive any email yet, will confirm when received.

myounus96 commented 4 years ago

Got the email.

JACKYLUO1991 commented 4 years ago

@ success or failure?

myounus96 commented 4 years ago

@ success or failure?

Getting this error FileNotFoundError: [Errno 2] No such file or directory: './segmentation/ParseNet_120_G.pth'. By placing trained model in segmentation folder, getting this size mismatch for classifier2.weight: copying a param with shape torch.Size([19, 512, 1, 1]) from checkpoint, the shape in current model is torch.Size([19, 384, 1, 1]).. @JACKYLUO1991

JACKYLUO1991 commented 4 years ago

@myounus96 You need to train the segmentation network first. What you were given before is the weight of face detection.

myounus96 commented 4 years ago

@myounus96 You should first configure the relevant parameters in parameters.py, and then run the main.py script. Of course, first you need to train the model...

Have done that..tester.py only producing accuracy metrics..not coloured masked predicted images..

Already done that, have mentioned in my 2nd comment. ^

JACKYLUO1991 commented 4 years ago

@myounus96 I think line 390 of the detection.py script is what you need.