Dynamic Serpentine Convolution with Attention Mechanism Enhancement for Beef Cattle Behavior Recognition

Animals, Journal Year: 2024, Volume and Issue: 14(3), P. 466 - 466

Published: Jan. 31, 2024

Behavior recognition in beef cattle is a crucial component of behavior warning and intelligent farming. Traditional faces challenges both difficulty identification low accuracy. In this study, the YOLOv8n_BiF_DSC (Fusion Dynamic Snake Convolution BiFormer Attention) algorithm was employed for non-intrusive behavior. The specific steps are as follows: 45 were observed using fixed camera (A LINE OF DEFENSE) mobile phone (Huawei Mate20Pro) to collect filter posture data, yielding usable videos ranging from 1 30 min length. These cover nine different behaviors various scenarios, including standing, lying, mounting, fighting, licking, eating, drinking, walking, searching. After data augmentation, dataset comprised 34,560 samples. convolutional layer (CONV) improved by introducing variable convolution dynamic snake-like modules. convolution, which yielded best results, expanded model’s receptive field, dynamically perceived key features behavior, enhanced algorithm’s feature extraction capability. Attention mechanism modules, SE (Squeeze-and-Excitation Networks), CBAM (Convolutional Block Module), CA (Coordinate Attention), (Vision Transformer with Bi-Level Routing introduced. attention mechanism, selected its optimal performance, ability capture long-distance context dependencies. computational efficiency through query-aware perception. Experimental results indicated that achieved among all algorithms terms accuracy, average precision at IoU 50, 50:95. accuracy reached 93.6%, 50 50:95 being 96.5% 71.5%, respectively. This represents 5.3%, 5.2%, 7.1% improvement over original YOLOv8n. Notably, recognizing lying 98.9%. conclusion, demonstrates excellent performance high-level fusion, displaying high robustness adaptability. It provides theoretical practical support management cattle.

Language: Английский

---

Programming and Setting Up the Object Detection Algorithm YOLO to Determine Feeding Activities of Beef Cattle: A Comparison between YOLOv8m and YOLOv10m

Animals, Journal Year: 2024, Volume and Issue: 14(19), P. 2821 - 2821

Published: Sept. 30, 2024

This study highlights the importance of monitoring cattle feeding behavior using YOLO algorithm for object detection. Videos six Charolais bulls were recorded on a French farm, and three behaviors (biting, chewing, visiting) identified labeled Roboflow. YOLOv8 YOLOv10 compared their performance in detecting these behaviors. outperformed with slightly higher precision, recall, mAP50, mAP50-95 scores. Although both algorithms demonstrated similar overall accuracy (around 90%), reached optimal training faster exhibited less overfitting. Confusion matrices indicated patterns prediction errors versions, but showed better consistency. concludes that while are effective behaviors, superior average performance, learning rate, speed, making it more suitable practical field applications.

Language: Английский

---

Field Implementation of Precision Livestock Farming: Selected Proceedings from the 2nd U.S. Precision Livestock Farming Conference

Animals, Journal Year: 2024, Volume and Issue: 14(7), P. 1128 - 1128

Published: April 7, 2024

Precision Livestock Farming (PLF) involves the real-time monitoring of images, sounds, and other biological, physiological, environmental parameters to assess improve animal health welfare within intensive extensive production systems [...]

Language: Английский

---

Computer vision algorithms to help decision-making in cattle production

Animal Frontiers, Journal Year: 2024, Volume and Issue: 14(6), P. 11 - 22

Published: Dec. 1, 2024

Language: Английский

---

Cattle Anomaly Behavior Detection System Based on IoT and Computer Vision in Precision Livestock Farming

Published: Dec. 12, 2024

Language: Английский

---