An Application of Pixel Interval Down-Sampling (PID) for Dense Tiny Microorganism Counting on Environmental Microorganism Images

This paper proposes a novel pixel interval down-sampling network (PID-Net) for dense tiny object (yeast cells) counting tasks with higher accuracy. The PID-Net is an end-to-end convolutional neural (CNN) model encoder–decoder architecture. operations are concatenated max-pooling to combine the sparse and features. addresses limitation of contour conglutination objects while counting. evaluation was conducted using classical segmentation metrics (the Dice, Jaccard Hausdorff distance) as well metrics. experimental results show that proposed had best performance potential tasks, which achieved 96.97% accuracy on dataset 2448 yeast cell images. By comparing state-of-the-art approaches, such Attention U-Net, Swin U-Net Trans can segment clearer boundaries fewer incorrect debris, shows great in task accurate