@Mohamed I seem to be having a similar issue with an object detection dataset. It was fully labelled in RoboFlow and the labels look fine there. But, when I expect the images and labels to do some training with yolov7, the bounding boxes no longer match what was on RoboFlow. Auto-Orient doesn’t seem to fix the issue for me. Appreciate your support resolving this. Thanks!
Outputs per training example: 3
Crop: 0% Minimum Zoom, 25% Maximum Zoom
Rotation: Between -10° and +10°
Shear: ±10° Horizontal, ±10° Vertical
Hue: Between -15° and +15°
Saturation: Between -15% and +15%
Brightness: Between -25% and +25%
Blur: Up to 2px
Noise: Up to 1% of pixels
I am still having this issue for an object detection dataset using augmentation. Any tips? Thanks
Has this been solved? I’ve noticed my object detection dataset + augmentation produces different results outside of roboflow (boxes are not correct).
Has this been resolved? I am having the same issue. I tried adding the Auto Orient pre-processing step but same issue.
I’ve merged all your replies since they were regarding the same issue. Did you try generating a version without auto orient?
Hi yes that is how I did it originally. Then, based on suggestions from the forum, I tried with auto-orient but the issuer persists.
Can you share the ID of your workspace and project, or a link to the project on Universe?
BERKELEY LAB, SORMA v4 and v5
Apologizes for the late response. As a troubleshooting step: can you try applying Auto-Orient with a resizing option to see if you still experience the issues you are referring to?