In commercial floor-rearing environments for yellow-feathered broilers, visual screening of broilers with observable health-related abnormalities is challenged by low illumination, individual occlusion, and the limited use of local comb information. This study proposes a two-stage global–local framework for yellow-feathered broilers. Images were collected from one commercial floor-rearing farm; after screening, 1463 images were used for key-region localization, and 2031 samples were constructed for health-status recognition. Samples were labeled as Healthy or Unhealthy by expert consensus according to observable phenotypic characteristics, rather than veterinary-confirmed disease diagnoses. First, a Dual-Target YOLO (DT-YOLO) model localized whole-body and comb regions. Second, a Transformer network integrated features from paired or single available inputs for classification. DT-YOLO achieved a mean Average Precision (mAP) of 97.56% at the 0.5 IoU threshold. For paired whole-body–comb inputs, the proposed classification model achieved an Accuracy of 97.11%, Macro-Precision of 97.38%, Macro-Recall of 96.58%, and Macro-F1 of 96.95%. Under the mixed-input condition, the model maintained an Accuracy of 96.06% and a Macro-F1 of 96.02%. These findings indicate that combining whole-body and comb information can support visual screening of unhealthy broilers under the single-farm floor-rearing conditions represented in the current dataset. Its applicability to other farms, breeds, and management conditions requires independent external validation.

Automated monitoring of broiler feeding can improve management, but occlusion and scale variation hinder reliable detection in commercial multi-tier cages. This study developed a lightweight vision model for feeding behavior detection (RE-FB) and a feeding activity level (FAL) indicator for group-level assessment. The 2300 annotated images of yellow-feathered broilers were divided 8:1:1 for training, validation, and testing. RE-FB combines reparameterized multi-branch feature extraction, attention, and bidirectional multi-scale fusion to detect feeding birds and segment cages and feed troughs. On the test set, RE-FB-m achieved 89.7% precision, 87.3% recall, 92.1% box mAP@0.5, and 99.6% mask mAP@0.5, with 8.1 million parameters and a reported model inference speed of 158 FPS on the test platform. RE-FB-m achieved the highest box mAP@0.5 for feeding behavior detection among the evaluated Faster R-CNN, RT-DETR, and YOLO models. RE-FB was then applied to videos of four fixed groups monitored from 11 to 17 days of age. FAL was defined as the time-averaged proportion of the total birds in each monitored group detected as feeding, expressed as a percentage. Age-specific mean FAL values across the four groups ranged from 7.95% to 8.74%, with no significant age effect detected (p = 0.175). After controlling for group effects, the association between FAL and same-day weight gain was not statistically significant (partial r = −0.106, p = 0.614). RE-FB accurately and efficiently detects feeding behavior in caged broilers and generates group-level FAL records by aggregating detections over time. This approach provides a technical basis for non-contact monitoring of feeding activity in broiler production and can provide quantitative behavioral information for feeding management.