Unable to convert weights file into tf

[cat768]

my custom weights file is junglecamp0.6.weights, based on yolov3 (not yolov3-tiny)
31 classes

When I run the command " python load_weights.py --weights ./weights/junglecamp0.6.weights --output ./weights/junglecamp0.6.tf " it gives me the following error message:

ValueError: cannot reshape array of size 76070 into shape (256,128,3,3)

[cat768]

[cat768]

Here is my cfg file:

junglecamp0.5.txt

(saved it to a txt file instead of cfg because github won't allow me to upload cfg)

[safayildirim]

You should provide number of custom classes you have as parameters.

[cat768]

You should provide number of custom classes you have as parameters.

How do I do that? (I have 31 classes)

[ejri]

in the load_weights.py file, change the number of classes from 80 to 31 in your case.

Also make sure to make the changes in app.py and detect.py

[Shreyas0812]

I did this, but got the same error...

Object-Detection-API/yolov3_tf2/utils.py", line 68, in load_darknet_weights
conv_shape).transpose([2, 3, 1, 0])
ValueError: cannot reshape array of size 209204 into shape (512,256,3,3)

[Zwj321123]

in the load_weights.py file, change the number of classes from 80 to 31 in your case.

Also make sure to make the changes in app.py and detect.py

This works for me! Thank you!