Deep Learning Approaches for Livestock Monitoring: Comparing YOLOv11 and Faster R-CNN for Chicken Detection

Object detection is essential for the progress of smart farming systems. This research offers a comparative examination of two advanced deep learning models, YOLOv11 and Faster R-CNN, for chicken detection utilizing the publicly accessible "Chickens-Oussaifi" dataset. The dataset contains 5,490 labeled pictures of chickens taken from several places, such as pictures from the internet, screenshots from games, and frames from videos. Standard object detection metrics were used to train and test both models in the same computer environment. With a mAP@50 of 0.867, a precision of 0.788, a recall of 0.819, and a fast inference time of 2.3 milliseconds per image, YOLOv11 was very good at finding things. Faster R- CNN had a higher recall rate (0.9135), but a lower precision rate (0.5312) and a slower inference time (80.6 milliseconds per image). These results show the pros and cons of each model, which helps choose the best detection frameworks for real-time intelligent poultry monitoring and future use in larger smart farming applications.