A lightweight defect grading detection method for quick-frozen dumplings based on improved YOLOv11n
Year: 2026
Type: article
Abstract: Fine-grained defect grading and differentiated quality management of quick-frozen dumplings remain challenging in complex industrial environments. To support fine-grained quality evaluation, a defect grading dataset containing four quality levels (normal, minor defect, moderate defect, and severe defect) was constructed. Based on this dataset, a lightweight defect grading detection model, SCSL-YOLO, based on YOLOv11n is proposed. StarNet and the C3k2_Star module are introduced into the backbone network to enhance multi-scale feature representation. A cascaded group attention mechanism (C2CGA) is adopted in the feature fusion stage to improve feature discrimination under complex backgrounds. In addition, a lightweight shared detail-enhanced detection head (Detect_LSDECD) is designed by integrating detail-enhanced convolution and group normalization to improve bounding box prediction and fine-grained quality classification while reducing computational complexity. Experimental results show that SCSL-YOLO achieves superior detection accuracy and efficiency compared with YOLOv11n. Precision, recall, and [email protected] reach 95.7%, 94.5%, and 98.6%, respectively. Compared with the baseline model, precision is improved by 2.4 percentage points, while recall and [email protected] are increased by 1.3 percentage points and 0.5 percentage points, respectively. Meanwhile, the number of parameters, model size, and GFLOPs are reduced by 40.38%, 30.91%, and 25.40%, respectively, and the inference speed reaches 316.6 FPS. The proposed method achieves a favorable balance between detection accuracy and computational efficiency, providing an effective solution for frozen dumpling defect grading detection and intelligent food quality inspection.
Source: Scientific Reports
Authors: Du Xiong, Yuping Wang, Baoyong Guo, Xinming Ma
Language: English
Subfield: Industrial and Manufacturing Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $2,690
OpenAlex ID: W7237355254
FWCI:
Cites:
Cited by:
Related to:
