Automatic Segmentation and Detection of Apples using Computer Vision

Apple detection and segmentation are important functions in agricultural automation and food processing industry. Accurate detection and segmentation of apples from images or videos will help in many applications such as yield estimation, quality assessment, and harvesting. In this paper, we propose a deep-based apple detection and segmentation system using state-of-the-art convolutional neural networks (CNN) and image segmentation techniques. Our approach includes the first step of image enhancement followed by generation of regional proposals using a CNN-based detection model. The segmentation network then adjusts the detected regions to accurately identify the boundaries of the fruits. We evaluate our method on different datasets of fluid images taken in different environments and plant areas. Experimental results demonstrate the effectiveness of our method with an average precision (AP) of 0.78, recall of 0.82, F1 score of 0.80 and intercept (IoU) of 0.85. These results demonstrate the robustness and accuracy of our method in detecting and classifying apples.