# How to increase accuracy of image orientation classification (Left, Right, Center)?

I am working on classifying images in "Left", "Right", "Center", "Back". Training and Validation images look like this:

The images are "Left", "Right", and "Center". I am following Pytorch transfer learning tutorial with Resnet50 architecture and have not changed anything.

The transformations I am using is as follows.

data_transforms = {"train"  : A.Compose([
A.Resize(256,256),
#  A.ShiftScaleRotate(shift_limit=0.05, scale_limit=0.05, rotate_limit=15, p=0.5),
#  A.RandomCrop(height=128, width=128),
A.RGBShift(r_shift_limit=15, g_shift_limit=15, b_shift_limit=15, p=0.5),
A.RandomBrightnessContrast(p=0.5),
A.Normalize(mean=(0.485, 0.456, 0.406), std=(0.229, 0.224, 0.225)),
ToTensorV2(),]),
"val": A.Compose([
A.Resize(256,256),
# A.CenterCrop(height=128, width=128),
A.Normalize(mean=(0.485, 0.456, 0.406), std=(0.229, 0.224, 0.225)),
ToTensorV2(),])
}


Initially I had back images too, but they were getting miss-classified with center a lot with val accuracy 65%.

Without back images I got 74% accuracy with labels left, right and center. Modified it to classify only center and not center, with only three type of images "left", "right" and "center" I have achieved 91% of val accuracy.

I am looking for ways to increase accuracy for classifying images in left, right, center.

• Hi sandepp! does your dataset contain only one object? Sep 13 at 23:43
• No, there are various products which I gathered from different sources mostly include electrical products like these. One of the sites I scrapped was Eaton Sep 14 at 11:18