Low Test Accuracy

#1

I have implemented both alexnet and mobilenet. In mobilenet my training accuracy is 90% while testing accuracy is just 5%.

Here is the link to my code:

https://drive.google.com/file/d/1RoXIFkrV2NKEF-6JV6YgK2dSHdyhSfjL/view?usp=sharing

#2

hey @Varunsh_20 ,
There are somethings you need to work upon

  1. Take large number of different augmentations when you need to create more data . Currently you are not creating new data , you are just using augmentations to augment you data at the training time.

  2. Don’t augment validation and testing data. As you dont know what augmentations it can take as they are being taken at random. And even a slight change in your input image your output can change a lot.

  3. Your dataset is very small , so try making a custom small model on your own.

  4. Dont provide validation steps and steps per epoch , while using generators it will decide those steps automatically.

  5. Try with different batch size like 16 ,64.

Try these points i hope they will help you achieve better.
Thank You and Happy Learning :slightly_smiling_face:.

#3

I hope I’ve cleared your doubt. I ask you to please rate your experience here
Your feedback is very important. It helps us improve our platform and hence provide you
the learning experience you deserve.

On the off chance, you still have some questions or not find the answers satisfactory, you may reopen
the doubt.