lijiannuist / lightdsfd

light DSFD

lightDSFD

By Jian Li

Introduction

We propose lightDSFD based on DSFD. the original DSFD code is here.

On Nvidia Tesla P40，the time of network is 13ms.

模型 NMS_Params	Easy Set	Medium Set	Hard Set
lightDSFD,thresh=0.05,number=500	0.891	0.864	0.469

DSFDv2

We update DSFD to generate DSFDv2, welcome to use. Here is weights

Contact

QQ群：1078945274

lightdsfd's Issues

TypeError: forward() missing 1 required positional argument: 'prior_data'?

Error code:
output = self.detect(
face_loc.view(face_loc.size(0), -1, 4), # loc preds
self.softmax(face_conf.view(face_conf.size(0), -1, self.num_classes)), # conf preds
self.priors.type(type(x.data)) # default boxes
)

Error message:
Traceback (most recent call last):
File "C:/Users/JUICE/Desktop/lightDSFD/test.py", line 215, in
light_test_oneimage()
File "C:/Users/JUICE/Desktop/lightDSFD/test.py", line 180, in light_test_oneimage
det = infer(net , img , transform , thresh , cuda , shrink)
File "C:/Users/JUICE/Desktop/lightDSFD/test.py", line 99, in infer
y = net(x) # forward pass
File "D:\ProgramData\Anaconda\envs\pytorch12\lib\site-packages\torch\nn\modules\module.py", line 547, in call
result = self.forward(*input, **kwargs)
File "C:\Users\JUICE\Desktop\lightDSFD\light_face_ssd.py", line 334, in forward
self.priors.type(type(x.data)) # default boxes
TypeError: forward() missing 1 required positional argument: 'prior_data'

few faces detected

few faces gets detected by running test.py. Does it requires any parameter changes during inference to detected all faces

?

How implement this model with camera?

I want to use your model to detect face with camera. How i can do that? If someones know , tell me plz. Thank you.

ValueError: not enough values to unpack (expected 2, got 0)

Hi, thank you for sharing the code.
When I run test.py with light_test_widerface(), I get the following error:

Traceback (most recent call last):
  File "test.py", line 215, in <module>
    light_test_widerface()
  File "test.py", line 207, in light_test_widerface
    det = infer(net , img , transform , thresh , cuda , shrink)
  File "test.py", line 98, in infer
    y = net(x)      # forward pass
  File "/home/v-chenqy/.pyenv/versions/py37/lib/python3.7/site-packages/torch/nn/modules/module.py", line 532, in __call__
    result = self.forward(*input, **kwargs)
  File "/home/v-chenqy/Repo/lightDSFD/light_face_ssd.py", line 333, in forward
    self.priors.type(type(x.data))                 # default boxes
  File "/home/v-chenqy/Repo/lightDSFD/layers/functions/detection.py", line 76, in forward
    ids, count = nms(boxes, scores, self.nms_thresh, self.top_k)
ValueError: not enough values to unpack (expected 2, got 0)

could you support the code for train?

DSFDv2 and light DSFD

Hi, is the DSFDv2 also a light version of DSFD or is it just an upgrade of DSFD?

could u add a requirements.txt?

I got a lot of error runnint test.py in torch==1.7