search

g-h-anna / ultralytics4channel

Ultralytics 8.2.5 modified to train and predict for 4-channel (RGBD) images.
GNU Affero General Public License v3.0
11 stars 0 forks source link

predict issue #3

Open 13253318778 opened 1 month ago

13253318778 commented 1 month ago

Hi,thank you very much for the outstanding work you have done with yolov8RGBD. Based on your work, I completed the training process, but I encountered some problems during prediction. The images I predicted did not generate any bounding boxes or anything like that; the images look just like the input images. Have you experienced the same problem?
My predicted output images are as follows, and after the prediction is completed, the prompt still indicates that it is 3-channel. 20240310_457 6eb824a6e79efdb3344f284398ed829

Kedharkb commented 1 month ago

Hi @13253318778, i am also facing the above issue, were you able to fix it?

13253318778 commented 1 month ago

Hi @13253318778, i am also facing the above issue, were you able to fix it?

Hi, I apologize, I am still working on it and haven't resolved the issue yet.

g-h-anna commented 1 month ago

Hi,thank you very much for the outstanding work you have done with yolov8RGBD. Based on your work, I completed the training process, but I encountered some problems during prediction. The images I predicted did not generate any bounding boxes or anything like that; the images look just like the input images. Have you experienced the same problem? My predicted output images are as follows, and after the prediction is completed, the prompt still indicates that it is 3-channel. 20240310_457 6eb824a6e79efdb3344f284398ed829

Hi @13253318778 ,

I didn't experience this type of error. Can you show the output log of the training?

13253318778 commented 1 month ago

Hi,thank you very much for the outstanding work you have done with yolov8RGBD. Based on your work, I completed the training process, but I encountered some problems during prediction. The images I predicted did not generate any bounding boxes or anything like that; the images look just like the input images. Have you experienced the same problem? My predicted output images are as follows, and after the prediction is completed, the prompt still indicates that it is 3-channel. 20240310_457 6eb824a6e79efdb3344f284398ed829

Hi @13253318778 ,

I didn't experience this type of error. Can you show the output log of the training?

Thank you very much for your reply. Here is some information about the process of training my model. As you can see, the images are indeed four-channel. Is this the information you need? ![Uploading 微信图片_20240530204115.png…]()