LFD-CD: Peripheral Blood Cells Detection Using a Lightweight Cell Detection Model with Full-Connection and Dropconnect.

Blood testing is an important basis for the diagnosis of many diseases. However, the accuracy of detection instruments in hospitals is often not high, and manual detection of blood samples is sometimes necessary. The whole detection process is not only time-consuming, but also limited in terms of the types of detection. The addition of Deep Convolution Neural Network (DCNN) can quickly and accurately detect blood cells. This study designs a Lightweight Cell Detection Model with Full-connection and Dropconnect (LFD-CD) to achieve the detection task of the peripheral blood cell dataset. In the detection of 8 categories, the model is optimized by adding the full connection layer and Dropconnect model, and the accuracy of the LFD-CD results achieves 99.3%. In the comparative experiment, LFD-CD demonstrates higher detection accuracy and faster detection speed than other independent DCNN models. Moreover, the space required for LFD-CD is only 395 MB, which is half the size of the YOLO-v7 model used in the comparison experiment.