立项:有人爱下棋却很菜,在对战过程中又不愿被菜,利用这个辅助系统可以轻松战胜公园大爷
受众:象棋爱好者
点击空白处退出提示
立项:有人爱下棋却很菜,在对战过程中又不愿被菜,利用这个辅助系统可以轻松战胜公园大爷
受众:象棋爱好者
1.AI识别盘面
2.AI(象棋鱼)支招
The application follows a complete data-to-decision pipeline:
Floating UI A custom WindowManager-based overlay for non-intrusive user interaction.
Screen Capture Uses the MediaProjection API to acquire real-time frames without modifying the target game.
Image Processing
Dynamic chessboard detection and cropping
Grid alignment and segmentation into 90 individual square cells
CNN Inference (On-device)
A lightweight CNN (PyTorch → TFLite) classifies 14 piece types + empty
Optimized for mobile CPU inference
FEN Generation
Converts visual board state into Forsyth–Edwards Notation (FEN)
Engine Communication
Sends FEN to the PikaFish (UCCI) engine via a background service
Result Overlay
Displays best move and engine evaluation on a simplified mini-board
1.盘面识别生成串码给象棋鱼引擎
2.象棋鱼生成支招串码
3.小棋盘显示支招
细节见:https://github.com/coder4world/Android-ChessAI-Assistant


评论