I have tried a lot to converge my model with 9053 samples, but loss value doesn't go lower than 2.3.
I have observed the lower number of samples the more loss goes down.
For example when I try to over fit on 320 samples it reaches to 0.4 and when do on 32 images it reaches almost to zero!
Please let me know if anyone has been successful in achieving satisfying outcome.
I have tried a lot to converge my model with 9053 samples, but loss value doesn't go lower than 2.3. I have observed the lower number of samples the more loss goes down. For example when I try to over fit on 320 samples it reaches to 0.4 and when do on 32 images it reaches almost to zero! Please let me know if anyone has been successful in achieving satisfying outcome.
Thanks