This work is an oil particle detection device based on image processing. Through the collection and processing of oil particle images, the device can distinguish and count oil particles. Specifically: 1. The oil is extracted from the test tube by a peristaltic pump (replaceable), and the oil flows into the circulation pool through the pipeline. 2. CCD camera and micro microscope collect pictures in the flow cell and transmit them to Jeston Nano TX2 (a development kit with GPU produced by Avida) chip for subsequent processing. 3. The chip performs algorithm processing (target detection, classification, segmentation) on the transmitted image to obtain the detailed information of the oil particles and display it on the screen. 4. The chip uploads the collected pictures, videos and particle information to the cloud server through the antenna for storage. This project can be divided into three parts: device structure, back-end development and front-end development. 1. Device structure The housing of the device is mainly composed of five sheet metal housings, a 3D printed base and various 3D printed components. The interior includes a 220V to 5V DC power supply, an adjustable micro DC peristaltic pump, a flow cell, an oil pipeline, a voltage regulating circuit, a micro microscope, a CCD camera, a lighting power supply, a display screen, etc. 2. Back-end development The server of this project is a lightweight Web server based on Linux. The server can support a relative number of concurrent clients and respond in time, store user information, realize user login and registration functions, and support client access to information such as pictures and videos in the server. After webbench stress testing, tens of thousands of concurrent amounts can be realized, tens of thousands of QPS can be run locally, and about 1000 QPS on the server. the backend code includes thread pools, timers, logging systems, and mysql connection pools. 3. Front-end development There are four modules in the front end, which are to turn on the camera, start recording, save pictures and upload to the server. Click the "Open Camera" button to open the camera; Click the "Start Recording" button to record video; Click the "Save Video" button to save the video; Click the "Save Picture" button to save the picture; Click the "Upload to Server" button to upload the current video, picture and oil information to the server.