Based on the current mainstream pathological staining methods, this paper studies and designs a scheduling method applied to the pathological autostainer process. The staining process performed by the pathological autostainer is based on clinical pathological standards, and a series of continuous staining operations are carried out on pathological slices. The operation process should follow a fixed staining stage. The pathologically required stained slices were strictly immersed in the corresponding reagents to achieve the demand of staining dehydration. According to the characteristics of different reagents, the staining process has high requirements in the staining time accuracy of pathological slices. The scheduling method in this paper builds a mathematical model of the staining process of the autostainer according to the time characteristics of the staining process, and uses the Simulated Annealing Algorithm to obtain the minimum total staining time under the input of multiple slices. Through the simulation of experimental data, the algorithm was analyzed and proved to be effective for the pathological staining process.
To address this issue that manual detection of animal fiber is subjective and requires experienced inspectors, an animal fiber detection algorithm GTB-YOLOv8 based on YOLOv8 is raised. In this paper, a new kind of C2f module is put forward by using convolution method of GhostnetV2. To enhance character extracting, Triplet Attention is introduced into the Neck section. PANet is transformed into BiFPN for more efficient extraction of contextual information. To increase the rate of convergence and the precision of regression, a new method called Wise-IOU is proposed, which could be used as a substitute for original CIOU. Through ablation experiments and model comparison experiments, the GTB-YOLOv8 algorithm is shown to have good performance on the wool cashmere data set : the [email protected] of the GTB-YOLOv8 model has been enhanced from 89.3% to 92%, resulting in a remarkable increase of 2.7%. Moreover, the model's parameter count and computational workload have been significantly reduced by 28.9% and 25.9%. Compared to other classic methods of detection, GTB-YOLOv8 has obvious benefits in the detection of wool and cashmere. In the future, we aim to apply the animal fiber detection algorithm to plant fiber detection, so that the model can achieve higher accuracy and have stronger robustness.