AI Bet

Inspiration

What it does

Our program is able to collect statistics such as: match predictions and match betting odds from various providers from Rapid API. We combine and display this data on our program to assist betters in deciding which bets will maximise their return. An additional function of the program is that it can recognise famous football players.

How we built it

Bet Assistanse: We use API Football provided by RapidAPI to collect the necessary data required for our program. We collect both the predictions and odds made by each bookie for a fixture and combine this data to estimate which bet will potentially provide the best returns. This program is built in python using the requests library to receive the data required. The data is parsed using json. The information is displayed by a GUI created by a Python using Tkinter.

Face recognition program: We trained a machine learning model using Teachingmachine. We collected about 20 training images for each individual football player. Then we use OpenCV and TensorFlow to identify them. Then we test the program by displaying players from our phone to our computer webcam.

x --> find about 30 picture run tensor flow TeachingMachine -->open CV, tensor flow

Challenges we ran into:

AI goal scoring identifier: We tried to train a model to identify goal scorings: Did not work because of small training samples and it is hard for the computer to identify the relation between the ball, player and the goal. We had to discontinue this side challenge due to time constrains.

Face Recognition program: We had a hard time having accurate face recognition. We at last found TensorFlow to solve the problem. However TensorFlow is M1 chip unfriendly and, so, our MacBook cannot run the program. By chance we found that the moon library allow us to run the program on our MacBook.

Bet Assistance: There is a limit of 100 request on API a day. At first we each create an account and we had 400 request a day. However we found out that each request will display insufficient information that we cannot request data frequently. So we at list out the relevant data and try to request data efficiently and store the data into our computer.