Rapid and Accurate Crayfish Sorting by Size and Maturity Based on Improved YOLOv5

Abstract

In response to the issues of high-intensity labor, low efficiency, and potential damage to crayfish associated with traditional manual sorting methods, an automated and non-contact sorting approach based on an improved YOLOv5 algorithm is proposed for the rapid sorting of crayfish maturity and size. To address the difficulty in focusing on small crayfish, the Backbone is augmented with Coordinate Attention to boost its capability to extract features. Additionally, to address the difficulty in achieving high overall algorithm efficiency and reducing feature redundancy, the Bottleneck Transformer is integrated into both the Backbone and Neck, which improves the accuracy, generalization performance, and the model’s computational proficiency. The dataset of 3464 images of crayfish collected from a crayfish breeding farm is used for the experiments. The dataset is partitioned randomly, with 80% of the data used for training and the remaining 20% used for testing. The results indicate that the proposed algorithm achieves an mAP of 98.8%. Finally, the model is deployed using TensorRT, and the processing time for an image is reduced to just 2 ms, which greatly improves the processing speed of the model. In conclusion, this approach provides an accurate, efficient, fast, and automated solution for crayfish sorting.

Surround Sensing Technique for Trucks Based on Multi-Features and Improved Yolov5 Algorithm

“…The input of YOLOv5 is enhanced by mosaic data, which is the same as that of YOLOv4. The main principle is as follows: select a picture and three random pictures for random cropping, and then splice them as a training set of pictures into the neural network [23]. This can not only enrich the background of the data set, improve the robustness of the system, but also reduce the loss of GPU memory and accelerate the training speed of the network.…”

The Efficiency of YOLOv5 Models in the Detection of Similar Construction Details

“…For example, in medicine, image data can be used to predict different diseases, such as cancer [1,2], glaucoma [3,4], and pneumonia [5,6]. Object detection models can be used in systems for travel direction recommendation [7], in industry for solutions to robotization tasks [8,9], in face detection for different applications [10,11], or other fields [12][13][14][15][16]. Usually, in all research, various computer vision methods or combinations are used to solve the specific problem.…”