Hello, this is my first experience with 3d Hand Pose Estimation, I would like to know what is the difference between MSRA and other datasets, is it the different subjects？ And how to visualize the MSRA results since I have run python test_samples.py --dataset MSRA but it failed. Also, I'd like to know how to visualize the metrics(1.mean 3d error 2. fractions of the frame within distance) in the code.
Thank you very much.

After extracting MSRA dataset, everything is availabe but this file Data/MSRA/P1/5/joint.txt not found. Where can I find this file

Hello,

I am currently trying to implement a program that use Pixelwise Regression to estimate the pose of at least two hands on a depth video stream (one frame at the time).
I am using the Stereo Labs ZED Mini camera.
Since Pixelwise can only estimate the pose of only one hand on a frame, I begin by using Mediapipe Hands (I know it is overkill, I may change later) to locate the hands and crop them on the frame. Then I resize the cropped hands to a size of 128x128.
Finally I can use this code :

    def estimate(self, img):
        
        label_img = Resize(size=[self.label_size, self.label_size])(img)
        label_img = tr.reshape(label_img, (1, 1, self.label_size, self.label_size))
        mask = tr.where(label_img > 0, 1.0, 0.0)

        
        img = img.to(self.device, non_blocking=True)
        label_img = label_img.to(self.device, non_blocking=True)
        mask = mask.to(self.device, non_blocking=True)

        self.heatmaps, self.depthmaps, hands_uvd = self.model(img, label_img, mask)[-1]
        hands_uvd = hands_uvd.detach().cpu().numpy()
        self.hands_uvd = hands_uvd

        return hands_uvd

To get this

After I looked into the values of img, label_img and mask when executing test_samples.py, I got the feeling that contrary to mine those matrix are normalized, which could be the cause of my poor results.
So is my feeling right and if so can you explain to me how I can do the same treatment on my matrix.

P.S. : I tested with both HAND17 and MSRA pretrained models.

Thank you for your work.

When I look at the source code, it is difficult to understand why it is designed this way, colud you please explain it, thank you very much~
""""
min_d = max(du + dv - 1, 0)
max_d = min(du, dv)
d = (max_d + min_d) / 2
b = du - d
c = dv - d
a = 1 + d - du - dv

heatmap[low_v, low_u] = a
heatmap[low_v, low_u + 1] = b
heatmap[low_v + 1, low_u] = c
heatmap[low_v + 1, low_u + 1] = d"""

Hi there,
How can I get the pre-trained models for MSRA and ICVL?

Also, in the paper, it is mentioned that 10 epochs was used (with data aug), but the train code has 50 epochs set as default. Which one is correct?

Thanks

Hello, thank you for your amazing implementation.
Is this PixelwiseRegression repo were developed on windows 10 ?

Hello!How can I fix this Error as title?
I'm already download the NYU dataset and pretrain model(NYU_default_final.pt) but I confuse how can I get joint_data.mat?Thanks!

Hello, thanks for your work, in the paper you said ''Note that, we do not use XYZ coordinates as output like some models did, because the UVD coordinates are more direct information from depthimages without a transformation influenced by the intrinsicparameters of the camera'', so, when you compute the 3D error, you uesd UVD coordinates or XYZ coordinates ?

NameError: name 'img' is not defined！ The img on line 216 of the training code is not declared。How to solve it？

Hi there, can you please suggest how can i generate X_center_test.txt file to infer on a custom 2D hand image?