GingerGM is a chess company owned and run by Simon Williams, aka Ginger GM. Simon has a following of 47k subscribers on YouTube and streams regularly on twitch.com. [login to view URL] is currently selling chess courses available for streaming and download. In 2020, GingerGM is developing a web service for chess openings called gChess ([login to view URL]) aimed at professional, serious and hobby chess players.
We are now looking for a partner to developing some of the technology needed on the service. We are looking for someone with experience in AI, Machine Learning and image recognition systems.
Main Goal of the Software
gChess will have several features, among them the ability to put a position on a chess board and the system will instantly search among top games from tournaments, online sites ([login to view URL] and [login to view URL]), an encyclopaedia and video courses.
We will also include the ability to search chess channels on YouTube, so when a user puts a position on a board, the system automatically shows YouTube videos where the same position has occurred, with the ability to jump straight into the correct position in the video. To achieve this we need software that can parse a video file detecting any chessboards and extract the position as a FEN with a timestamp ([login to view URL]–Edwards_Notation).
The software must run on standard Linux distribution (no Windows). Language is optional. The interface is also optional, although a REST API would be preferred. Documentation on using the system from PHP should be included. This documentation should show we can give one YouTube video as input, and get a list of positions with timestamps as output. The accuracy of the system needs to be at least 98%.
The software must run on standard Linux distribution (no Windows). Language is optional. The interface is also optional, although a REST API would be preferred. Documentation on using the system from PHP should be included. This documentation should show we can give one YouTube video as input, and get a list of positions with timestamps as output.
Also check out:
[login to view URL]
[login to view URL]
[login to view URL] (the app does YouTube search as we want)
To qualify as an accepted project, the following criterias need to be fulfilled:
Software will run on a linux platform and will be written in Python.
Software will accept a local movie file as input and produce a list of chess positions in FEN with timestamps (seconds) as output.
Software will identify pieces with an accuracy level of at least 98%.
Software will be able to parse at least 10 images / second.
Software shall be delivered within 14 days from start of project.
Software should handle different board and pieces styles (see examples of what should be handled), and artifacts like arrows and symbols on the board.
Written documentation of basic usage of the Software
Hi, I am doing my Phd in computer vision and deep learning. I have also 3 years of experience in industrial image processing projects with python, OpenCV, Tensorflow and Keras. I would be happy to help you with this p Daha Fazla
Bu iş için 20 freelancer ortalamada $1506 teklif veriyor
Hi there! I have certain knowledge and experience in Python, ML and AI. I have done several project in image processing and object detection with Tensorflow and Keras. I think I can help you with my skills. Thanks.