8/8/2023 0 Comments Next best move chess bot![]() Obviously, there is still lots of room for improvement, but we are super proud of what we achieved in these twenty-four hours, as well as how it looks and feels. But, we managed to deliver a practically fully working application that connected all components that we originally planned. ![]() Definitely, a pleasant surprise for all of us, and we're very proud of how far we've gotten together in terms of learning, developing, and bonding! As it was most of our members' first hackathon ever, we didn't know what to expect by the end of the hackathon. Only one hour into the hackathon (while we were still trying to work out our idea), one of our members already had a huge component of the project (a website + active camera component + "capture image" button) as a rough draft. Also, they're so smart? They know so much!! Accomplishments that we're proud of Honestly, though, the only thing that got me through all this was my extremely supportive team!!! They were extremely understanding, supportive, and kind and I couldn't have asked for a better team. What the heck is an API? What's this? What's that? What are you doing right now? What is my job? What can I do to help? The infinite loop of questions kept on racing through my head. I had some Python and Java experience but now I realize there's a whole other world out there full of possibilities, opportunities, etc. Here's what Melody has to say about her personal learning curve: At first, it felt like I didn't know ANYTHING. Not to mention, splitting the work and especially connecting all components together was a challenge that we had to work extra hard to achieve. Machine learning, image recognition, computing languages, navigating through Github, are only some of the huge learning curves we had to overcome. It's pretty safe to say that all of us had to learn SOMETHING on the fly. We stayed resilient, optimistic, and determined to finish our project off with a bang! Learning Curves Each of us live in a different timezone and a large challenge was working around these differences. Whether if it's 8pm, 12am, 4am, it doesn't matter to us. We also trained the Deep Learning model with Keras and predicted the FEN string for the image taken from the webcam after image processing using OpenCV. To predict the next best move, Node.js (Express module) was used and stockfish.js was used to communicate with the most powerful chess engine so that we could receive information from the API to deliver to our user. Predicting next best move using FEN stream React.js and node.js were mainly used to create the website for our project (as it is a web app). NodeJS - To link front-end, DL model and Stockfishįigma was the main tool we used to design a prototype for the UI/UX page. Stockfish.js - The most powerful chess engine Keras - Deep Learning (DL) model to predict FEN string Kaggle - Data set which has 100,000 chess board images OpenCV - Convert image to an ortho-rectified chess board Front-end Technologyįigma - Used for prototyping the front end Some of our inspiration came from Apple's Camera app's ability to identify the URL of QR codes in an instant - without even having to take a picture. Run code through chess engine (we send the FEN code to stockfish (chess engine))Ĭhess engine will return next best move to us Turn position into text (by extracting the FEN code of it by using some machine learning model) ![]() Not only does it create an experience to help improve your own chess skills and strategies, but you can now analyze chessboards in real-time while watching your favourite streamers on Twitch. Have you ever wanted to see chess through the eyes of chess prodigy Beth Harmon? Where prodigies and beginners meet, BethtChess is an innovative software that takes any picture of a chessboard and instantly returns the next best move given the situation of the game. Between 2020 locking everyone indoors, and Netflix's Queen Gambit raking in 62 million viewers, everyone is either talking about chess, or watching others play chess. InspirationĢ020 has definitely been the year of chess. Some positions may be illegal such as both kings are under check. AcknowledgementsĬ for providing images of pieces and boards. Labels are in a filename in Forsyth–Edwards Notation format, but with dashes instead of slashes. Pieces were generated with the following probability distribution:Ģ Kings are guaranteed to be on the board. Images were generated using this custom-build tool Images were generated using 28 styles of chess boards and 32 styles of chess pieces totaling 896 board/piece style combinations. ![]() Contentġ00000 images of a randomly generated chess positions of 5-15 pieces (2 kings and 3-13 pawns/pieces) ![]() The goal of the project is to build a model able to generate FEN description based on a schematic image of a chess board. Get the next best chess move in real time from your camera! Context ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |