Abstract:

The aim of this study is to classify poultries using popular convolutional neural network models. The different YOLO models are experimented to find best YOLO models in terms of performance. For this purpose, a case study was conducted on different versions of the YOLO model. A new dataset has been described in this study. In the dataset, there are 918 photos containing chickens, cockerel, and chicks. The dataset split into %80 training set and %20 test set. The images of poultries in the training and test datasets were manually annotated and those in the training dataset were used to train the YOLOv3-tiny, YOLOv3, YOLOv4-tiny, YOLOv4, YOLOv5s, and YOLOv5x Models. The results of using YOLOv5 for poultry detection are compared with other popular CNN architectures, YOLOv3, YOLOv4 models. The results show that YOLOv5x (XLarge depth) model records the highest accuracy, resulting in a mean average precision at 0.5 IOU of %99.5

Keywords: