Technique used to detect small object in the image like shapes and number can used this technique in OCR.

Technique used when need to detect object from image but this is simple technique not advanced technique know to detect any objects in image very clear and this technique used to detect small things like number in image, shapes or things has speacific area.

1. To read image and tranform from rgb to gray.
2. Make threshold in image to be binary.
3. Using open-CV library.
4. Call connected_component from cv2.
5. Detect all shapes from the image.
6. Detect all number from sudoku image.
7. Deal with noise from the image to detect object only.

Before this you will be able to detect all object from the image without noise and I hope to add more details in connected component labeling about you know.