Saliency-Guided RT-DETR for Multi-Class Detection in Processing Tomato Sorting
TL;DR
SG-RTDETR is proposed, an RT-DETRv2 adaptation that combines detail-preserving downsampling, saliency-guided token encoding with residual spatial refill, context-aware feature organization, and adaptive cross-scale fusion for tomato sorting under controlled laboratory conditions.
Abstract
This study considers multi-class visual detection for processing tomato sorting under controlled laboratory conditions. In densely arranged images, occlusion and visual similarity can weaken the local boundary and texture cues needed to distinguish ripe tomatoes, unripe tomatoes, defective tomatoes, and soil clods. We propose SG-RTDETR, an RT-DETRv2 adaptation that combines detail-preserving downsampling, saliency-guided token encoding with residual spatial refill, context-aware feature organization, and adaptive cross-scale fusion. A four-class dataset was constructed from 1000 multi-object images and 400 single-object images used for supplementary representation learning. Across three independent training runs under controlled laboratory evaluation, SG-RTDETR achieved 87.8±0.4% mAP50:95, 92.4±0.4% mAP50, and 95.5±0.3% mAR50:95 (mean ± sample standard deviation). Relative to RT-DETRv2, the mean mAP50:95 increased by 3.4 percentage points, while total FLOPs remained at 61.17 G and forward-pass inference throughput decreased from 110.5 to 103.6 FPS. These results indicate an accuracy-oriented trade-off under the laboratory evaluation protocol; validation under realistic postharvest sorting conditions remains necessary.