jalalilabucla/phycv

PhyCV: The First Physics-inspired Computer Vision Library

phycv's Issues

pytorch 2.0

I was impressed by your talk during NVIDIA GTC. Congrats.
Is it possible to use PhyCV with current version of PyTorch i.e. PyTorch 2.0

tanks for your help

Extracting lines at specific angles using PAGE

Hello, may I know how can we extract lines that are at specific angles after getting output from page_gpu?

Example use-case is to determine the overall grain direction of an aluminum sheet.

Typo in pst.py

In method load_img:

if img_array is not None:
self.img = img_array
self.img = img_array.shape[0]
self.img = img_array.shape[1]

Missing paraemeters

Hello,,

The document says that there are two parameters for PST i.e. phase_strength and warp_strength. However,
in the pst.py I do not see these parameters. Am I missing something?

Regards

"PAGE" script returns a black image for GPU and a different output from CPU

Hi, I'm running the run_page.py on the attached images (with the original parameters).

With ring.jpg I get a completely black image from GPU (with original or any parameters). CPU output looks fine.

Downscaling the image, the GPU result is slightly different from the CPU one (and wrong in my opinion). I get a similar result cropping the original image a little bit.

Linux Mint 5.4.0-137-generic
Driver Version: 470.161.03
CUDA Version: 11.7.99
NVIDIA GeForce RTX 2070 with Max-Q Design

ring
smaller
smallest

Output on the "smaller" image (the image background is completely uniform):

CPU:

GPU:

Determining the best VEViD parameters

Hi, I was wondering what is the best way to determine the most suitable parameters while using VEViD for low-light enhancement. I see that the following was used in the examples. Is it a good idea to always leave them the same?

S = 0.2, T = 0.001, b = 0.16, G = 1.4

Add Arxiv article link to readme

Thanks for the software! If you wouldn't mind, adding the actual hyperlink to the Arxiv article will be nice. I mean in the Citation section of the readme. The link is:

https://arxiv.org/abs/2301.12531

cv2 is not import but used in pst_gpu.py

cv2 is not import but used in pst_gpu.py

Issue while working with opencv live camera feed

I am importing VEVID and VEVID GPU from phycv and cv2 in a single file.
After that I am trying to read live video feed using cv2 but there is no output. The camera starts but I do not see any output window when using cv2.imshow().

When I am commenting
from phycv import VEVID, VEVID_GPU
then I can use cv2 and its live feed otherwise I can not.
Any fix?