The sugar apple (Annona squamosa) is valued for its taste, nutritional richness, and versatility, making it suitable for fresh consumption and medicinal use with significant commercial potential. Widely found in the tropical Americas and Asia’s tropical or subtropical regions, it faces challenges in post-harvest ripeness assessment, predominantly reliant on manual inspection, leading to inefficiency and high labor costs. This paper explores the application of computer vision techniques in detecting ripeness levels of harvested sugar apples and proposes an improved deep learning model (ECD-DeepLabv3+) specifically designed for ripeness detection tasks. Firstly, the proposed model adopts a lightweight backbone (MobileNetV2), reducing complexity while maintaining performance through MobileNetV2′s unique design. Secondly, it incorporates the efficient channel attention (ECA) module to enhance focus on the input image and capture crucial feature information. Additionally, a Dense ASPP module is introduced, which enhances the model’s perceptual ability and expands the receptive field by stacking feature maps processed with different dilation rates. Lastly, the proposed model emphasizes the spatial information of sugar apples at different ripeness levels by the coordinate attention (CA) module. Model performance is validated using a self-made dataset of harvested optical images categorized into three ripeness levels. The proposed model (ECD-DeepLabv3+) achieves values of 89.95% for MIoU, 94.58% for MPA, 96.60% for PA, and 94.61% for MF1, respectively. Compared to the original DeepLabv3+, it greatly reduces the number of model parameters (Params) and floating-point operations (Flops) by 89.20% and 69.09%, respectively. Moreover, the proposed method could be directly applied to optical images obtained from the surface of the sugar apple, which provides a potential solution for the detection of post-harvest fruit ripeness.

Sugar apple (Annona squamosa) is prized for its excellent taste, rich nutrition, and diverse uses, making it valuable for both fresh consumption and medicinal purposes. Predominantly found in tropical regions of the Americas and Asia, its harvesting remains labor-intensive in orchard settings, resulting in low efficiency and high costs. This study investigates the use of computer vision for sugar apple instance segmentation and introduces an improved deep learning model, GCE-YOLOv9-seg, specifically designed for orchard conditions. The model incorporates Gamma Correction (GC) to enhance image brightness and contrast, improving target region identification and feature extraction in orchard settings. An Efficient Multiscale Attention (EMA) mechanism was added to strengthen feature representation across scales, addressing sugar apple variability and maturity differences. Additionally, a Convolutional Block Attention Module (CBAM) refined the focus on key regions and deep semantic features. The model’s performance was evaluated on a self-constructed dataset of sugar apple instance segmentation images captured under natural orchard conditions. The experimental results demonstrate that the proposed GCE-YOLOv9-seg model achieved an F1 score (F1) of 90.0%, a precision (P) of 89.6%, a recall (R) level of 93.4%, a mAP@0.5 of 73.2%, and a mAP@[0.5:0.95] of 73.2%. Compared to the original YOLOv9-seg model, the proposed GCE-YOLOv9-seg showed improvements of 1.5% in the F1 score and 3.0% in recall for object detection, while the segmentation task exhibited increases of 0.3% in mAP@0.5 and 1.0% in mAP@[0.5:0.95]. Furthermore, when compared to the latest model YOLOv12-seg, the proposed GCE-YOLOv9-seg still outperformed with an F1 score increase of 2.8%, a precision (P) improvement of 0.4%, and a substantial recall (R) boost of 5.0%. In the segmentation task, mAP@0.5 rose by 3.8%, while mAP@[0.5:0.95] demonstrated a significant enhancement of 7.9%. This method may be directly applied to sugar apple instance segmentation, providing a promising solution for automated sugar apple detection in natural orchard environments.