jubiiz / suguru

My take at a suguru solver

SUGURU SOLVER

This personnal project was meant to test my ability to build a big project from the ground up (from idea to result, with no premade youtube tutorial available). I also practiced using github and git in the process: all my coordination was done via those two tools.

Although it is far from user friendly, I am able to use it to solve 20x20 extra-hard sugurus in less than an hour. These usually take upwards of 5 hours by hand. Just for that, I consider it a success.

usage: The project takes as input an image of a suguru, then via many separate program files, the image is segmented into cases, the useful data is extracted from those images, then the suguru is represented in the code (as to be manipulated) and then solved. The solved version is displayed at the end.

The most important technologies it leverages are: CNNs: to extract number data cv2: for all image manipulation matplotlib: for solving process updates (lets the user see where it's at, useful for debugging) ac3 algorithm: for solving

suguru's Issues

CLI to help user-friendliness

A CLI could be used to navigate the different source-codes during the suguru resolution process (not only solving: processing, extraction, correction, solving)

Unoptimized CNN

The CNN used to extract numbers from the grid is not optimal: it cannot recognize empty cases:
Here's how it works:

1. reads the number in the case
2. if its a 1, check if the average of all the pixel is very low
3. if so, it's not a 1, it's a blank cell

number extraction needs supervision ; 4 not recognized

more resiliency of the neural network is needed. either different training data or different neural net
diff training data seems best rn imo

Non-square problems not supported

The code was not built to support non-square problems. Some parts (like in the processor.py automatic grid selection process) assume that the number of rows = the number of columns.

Inefficient manual grid ROI selection

Manual contour selection to find the ROI (the four corners of the grid) takes time and can be inefficient if done hastily. Consider adding an automatic feature so the program can do it for you (like with cv2.contour).

case crop slicing hard to read

At line 72 of pentomino.py, the slice to extract the individual case image is difficult to read.
The function of each term is not easily understood.
The issue could be resolved by relaying the cropping (or just slice production) task to a function.

Number extraction from cell image

Using the technology as used in CS50 AI to extract what number lies in each cell.
The cell images might have to be cropped again to remove any sidelines: they may interfere with the digit recognition.

More user-friendly code needed

The code has a number of user-unfriendly features that would benefit to be altered.
These features need to be corrected before issue #13 can be tackled.
Here is a brief list of what they are:

• multiple changes of filename required per source-file
• inconsistent filename input (sometimes it's "filename.txt", others just "filename")
• unclear and inconsistent data size input (a few files still rely on the fact that the problem will be 20x20, and do not have a centralized way of changing those numbers)

Case sides classification

From a single case image, extract which sides are bold and which are thin.