1.GitHub：https://github.com/wangshub/wechat_jump_game
2. Environment construction : The Python environment that has installed caffe and tensorflow frameworks can be used directly, install the ADB driver, and cmdinto theroot directory of wechat_jump_game-master to execute pip install -r requirements.txt to install the required environment. On the whole, the distance between the center point and the starting point of the chess piece is calculated through the screenshot, and adb simulates the realization process of the pressing time.
3. Enter the mobile phone and jump to the applet, connect to usb, execute adb devices in cmd to see if it is connected to the mobile device, enter the root directory of wechat_jump_game-master and execute python wechat_jump_auto.py
4. The test results are basically 500~800 points without parameter adjustment. Find the corresponding resolution of the model , modify the distance parameter in wechat_jump_game\master\config\1920x1080\config.json, change the parameter press_coefficient to 1.39, and calculate the center point of the piece in wechat_jump_auto.py. At the center point, several such pieces in a row are over. Debugging for dozens of minutes only to 2000 points.
