hy guys !!
I have an image classifier trained on cifar 10 dataset using tensorflow. its accuracy is 96.96%. The problem is that it classifies the images already in 32x32 size correctly but when the same image is passed to it in high resolution, it classifies it wrong.

I tried to make it resize the high resolution image to 32x32 and then read it but it gave wrong results. I also tried to resize the high resolution image to 32x32 and save it as .png and then read that image but it again classified it wrong.

I am frustrated as i dont know where the problem is and what should i do, please help if any of you can

0 Answers