How do I distinguish the stem from the rest of the onion body?

How to distinguish the stem from the rest of the onion for the input image below? The stem is the same color as the rest of the onion, the pointed portion (facing downwards) in the image attached below is the Stem of the onion.

This would need other user to share experience. See if there is appropriate model for this use-case. We have DeepStream SDK which includes demonstration of some models like Yolo and ResNet. You may give it a try and replace with the model.

Since the stem and the onion are of the same color, would semantic segmentation work in this case? Would semantic segmentation be able to tell things apart if they have the color?

You can prepare the data and train a model for the use-case. Maybe can use Yolo to do the training.

Using OpenCV functions, I’ve got to the stage where contours, inner and outer circles of the onion are drawn on the masked output.
Next I wish to rotate the entire onion such that the stem (present between the red and green circle in the image below) is facing vertically upwards.

Please suggest how I can go about doing this @DaneLLL

You would need to prepare more pictures and label the part in each picture. And do training with selected detection network to get the model. And apply it to DeepStream SDK.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.