search

WongKinYiu / yolov7

Implementation of paper - YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors
GNU General Public License v3.0
13.43k stars 4.23k forks source link

Is it possible for YOLO team to replace nn.conv2d with nn.conv3d in the future for medical imaging #1426

Open frabob2017 opened 1 year ago

frabob2017 commented 1 year ago

Is it possible for YOLO team to replace nn.conv2d with nn.conv3d in the future because it is very useful for medical imaging which is 3d format? I believe if it change to 3d level, it would be robust in medical imaging. I try to understand your code and do it. But it is beyond my capability.

faizan1234567 commented 1 year ago

YOLOv7 doesn't support 3D convolution; however, you can use 3D segmentation models on medical imaging. i.e. UNET or Encoder-decoder architecture. They work with 3D medical imaging.

frabob2017 commented 1 year ago

YOLOv7 doesn't support 3D convolution; however, you can use 3D segmentation models on medical imaging. i.e. UNET or Encoder-decoder architecture. They work with 3D medical imaging.

Hello faizan, thank you for your answer. Yes, from the programing point view, I found that UNET is easy to implement 3D convolution. Although more difficult for programming, YOLO can do it. Hope YOLO can implement this feature also.