-
-
Notifications
You must be signed in to change notification settings - Fork 3.4k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
RuntimeError: invalid argument 2: size '[16 x 3 x 15 x 13 x 13]' is invalid for input with 689520 elements at /pytorch/aten/src/TH/THStorage.cpp:84 #33
Comments
Please adjust the model accordingly for your dataset. The repository works for any object detection dataset, but you need to modify certain areas which are COCO-specific currently. The full list of directions is on this comment #6 (comment) |
@glenn-jocher Thanks! I've trained 68 epoches , finally the precision was 1, but recall was 0, did you met this trouble? need some advice. |
I had a similar problem, the best thing to do is to adjust the last constitutional layer before the [yolo] of the yolov3.cfg or the .cfg file that contains your model. |
This tutorial explains how to train custom datasets, including updating the yolov3.cfg file for your class count. |
@glenn-jocher @Electronicshelf @sporterman Hey guys!
I'm training the model on my custom dataset. I did follow the below steps.
I have got a doubt whether I should modify the classes=80 in the 3 [yolo] sections in yolov3.cfg file. I've also read this comment, but I don't know how to implement steps 4, 5, 6 in the above comment. UPDATE |
@Vysakhr great to hear its working for you now. We have updated the tutorials now to clearly state that the class counts need to updated in each YOLO layer in the *.cfg file, i.e. in your case from |
models.py
p = p.view(bs, self.nA, self.bbox_attrs, nG, nG).permute(0, 1, 3, 4, 2).contiguous()
when i use this coda run my own dataset, it stoped here , erros was as follow:
The text was updated successfully, but these errors were encountered: