Abstract: This study addressed the challenge of detecting the novel coronavirus (SARS-CoV-2) by developing a digital PCR droplet recognition system based on microfluidic chips and artificial intelligence algorithms. In this system, the microfluidic structures constructed utilizing semiconductor processing technology can precisely controll the generation of microdroplets, ensuring the uniformity and stability of droplets. By combining Gaussian blur technology and threshold segmentation algorithms, and using a pre-trained neural network model for intelligent recognition and counting of droplets, the detection sensitivity and accuracy are significantly improved. Experimental results show that the droplet recognition accuracy reached 99.41%, and the recognition accuracy of positive samples was 99.09%, demonstrating the potential application value of this system in point-of-care test scenarios.