Yolo Segmentation Chess
Project Flow
-
♔♕♗♘♙♚♛♝♞♟♖♜
-
The camera captures the image of chessboard then the image is analyzed using image processing to identify the moves made by an opponent and stockfish engine calculates the best possible move.
Method of Working
Step - 1
Image Transformation
Example
 |
 |
 |
 |
 |
| Input-Frame -> | Predicted Mask -> | Corner Points -> | Warp Perspective Image-> | Yolo Inference Image |
 |
 |
 |
 |
 |
- With help of unet model mask chess board and detect chess baord corner points with help of opencv operations. and apply warp perspective transformation on it.
| Input-Frame | warp-perspective | locate-boxes |
|---|---|---|
 |
 |
 |
Step - 2
| Image1 : Image of Chess Board befor player move piece | Image2 : Image of Chess Board after player move piece |
|---|---|
 |
 |
step - 3
| Difference of image by using function absdiff in CV2 | Change Difference_image to Gray scale image |
|---|---|
| diff = cv2.absdiff(image1,image2) | diff_gray = cv2.cvtColor(diff,cv2.COLOR_BGR2GRAY) |
 |
 |
step - 4
| Apply thresholding on Grayscale image | Find Contours on threshold image |
|---|---|
| matrix,thresold = cv2.threshold(diff_gray,value,255,cv2.THRESH_BINARY) | cnts,_ = cv2.findContours(thresold, cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_SIMPLE) |
 |
 |