What I did different - I used pip install bing-image-downloader instead of getting the bing API access, I’ve tried and I am unable to get the access for the API.
I am getting the following error - ValueError: This DataLoader does not contain any batches
Screenshot attached for reference:
I believe that this issue is what is causing my model to be trained with a continuous 50% error rate, attached screenshot below, please help, where am I going wrong.
@sha52 did you ever find out how to fix this issue? I am coming across this exact issue as well.
Thanks!
I think that part of the issue is that we are not downloading enough images. If you look at your confusion matrix, you are barely getting any images in your validation set. I couldn’t find the Azure key and I was using the DuckDuckGo downloader instead, but I can’t download more than 30 images.
It works for me know, I followed these instructions on how to download images from DuckDuckGo, and I made sure that I had at least 100 images per type. Downloading Bears using DuckDuckGo returns 0 images
