A CNN based image segmentation approach was used to identify fruit regions on the image. This was followed by a Watershed Segmentation to detect individual fruits in the scene.
Data taken at an apple orchard.
Image split from left to right:
- Raw image stream from orchard
- Segmentation output form classifier
- Threshold and morphological opening applied to segmentation output
- Watershed Segmentation used to identify individual fruits, which are then overlayed onto the original image.
