End-to-End Automated Crossword Solver using Image Processing, Natural Language Processing and Neural Network

Solving newspaper crosswords improves human brain function and mental health. It also helps decrease depression and anxiety that occurs due to social isolation and loneliness in the COVID-19 pandemic. In this game, players are enthusiastic about finding the correct answer after solving the puzzle. However, to check the answers, players have to wait until the newspaper publishes the solution. Due to this delay, players are moving to online crosswords. Mainly this issue occurs in Sri Lanka. There is no quick way to verify the answers using existing methods because players need the technical knowledge to use the solver. Also, existing solvers are unable to provide accurate solutions for different clue types. This paper provided an automated crossword puzzle solver which can solve British-Style crosswords using an image of the crossword puzzle. Image Processing, Natural Language Processing, and Neural Networking technologies were used to implement the proposed solution. The final answer list was prepared using the neural network model and filled the puzzle grid using OpenCV. The proposed system achieved more than 90% of word and letter accuracy in evaluation with human experts.