ABSTRACT
The demand for smart automatic system in postharvest technology, particularly in the postharvest of carrot production is high. In this chapter, a lightweight deep learning model (CDDNet) was constructed to detect surface defect based on ShuffleNet and transfer learning. Also, carrot grading methods were proposed based on minimum bounding rectangle (MBR) fitting and convex polygon approximation. Experimental results showed that the proposed CDDNet achieved a detection accuracy of 99.82% for binary classification (normal and defective) and 93.01% for multi-class classification (normal, bad spot, abnormity, fibrous root), and demonstrated good performance both in time efficiency and detection accuracy. The grading accuracy of MBR fitting and convex polygon approximation was 92.8% and 95.1%, respectively. This research provides a practical method for online defect detection and carrot grading, and has great potential in commercial packing lines.
