AI ML based app for cricket training For Machineroad
Executive Summary
About Client
Machineroad was started by Mitch Ferguson and Lockie Fergsuon both on top of thier cricketing skills and with the right knowledge and tools helping others in developing the game skills is what they wanted to do in Machineroad. With the mobile application goal was to help athletes to see how fast they can bowl and the areas for thier improvement. The competition in the sports sector is cut-throat and this app helps amateur as well as professional athletes to up their games.
Location: Auckland, New Zealand
Project - AI ML based mobile app for cricket training
Machineroad requirement was for implementing a bespoke AI ML based mobile app that helps to improve cricket bowling skills for their users. They wanted an app that helps their users to measure their bowling speed and creates a trajectory image snippet for the end user which further helps to understand the areas of improvement. Machineroad needed detailed analytics to help the users see their activities and compare results each week and month to help keep a track on the progress made. The requirement for AI ML based mobile app for cricket training was to be launched on both iOS and Android Store.
The Founder of MachineRoad Lockie Ferguson as world class cricket champion had this vision in mind ‘We want to bridge the gap between talent and success as a sportsman. Regardless of your upbringing we want you to be able to compete on the world stage and become the best athlete you can be”
Scope & Requirement
Below was the scope of work to develop a Cricket Training app:
- User should be able to download the app from Play and Google store if the device meets the specific requirement of camera and Video processing.Â
- User can then calibrate and start taking video when doing bowling and the app guides on the right placement and setup so as to get the most accurate video for processing and calculating the speed.Â
- AI and ML based video processing to give accurate results for the speed and if it there are issues like objects etc detected on the video it then informs the user that speed could not be calculated.Â
Implementation
Technology and Architecture
TechnologyÂ
The Mobile app was deployed with the below technological component
• Backend Code: .NET Core, C#, Node.js
• Mobile App code: Native Android. Native iOS
• Database: SQL Server, MongoDB
• Cloud: AWSÂ
Integrations
- Single Sign-on using Auth0
- Sendgrid for sending email notifications
- Single Sign-on using Auth0
Security:
• Data Encryption
• Multi-Factor Authentication for Admin, Teacher, and students when logging in
• All API endpoints are tokenized
Backup and Recovery
Cloud systems and components used in the attendance management system are secure and 99.99% SLA. We have added HA/DR mechanism to create a replica of the servicesÂ
Scalability
Application is designed to scale up to 10X times the average load received in the 1st 6 months of its usage and all cloud resources are configured for auto-scaling based on the load
Cost OptimizationÂ
Alerts and notifications are configured in the attendance management system to notify if the budget is being exceeded. Peritos being a cloud partner is managing the environment for the client keeping a close watch on the cost and finding ways to optimize the sameÂ
Code Management, Deployment
Code for the app is handed over to the client through Microsoft AppCenter.Â
CI/CD is implemented to automatically add, build and deploy any code changesÂ
Features of AI ML based Mobile app for cricket training
- It allows the users to create bowling videos and after the video is recorded it lets the user store the data and add to the player profile on the Machineroad app.Â
- The app records parameters like bowling speed, line, length and trajectory and saves the image and video for each sessionÂ
- It offers in detail analytics reports to compare weekly and monthly progress and an option to compare the performance with other users some of them being professional athletesÂ
- A monthly subscription that offers comparison charts of the monthly trainings and option to submit the speed and video for leaderboard.Â
- AL and ML based video processing to analyze the recorded videos which gives the speed same as compared to a speed gun.Â
- Integrated with social media and user can share training results on social media platforms including the badges and streaks they earned.Â
- We also implemented Gamification and Leaderboard functionality which motivates user to achieve higher results based on targets which can be customized for each user’s journey.Â
Challenges -AI ML based Mobile app
- Achieving similar results from one camera while the same results in an actual game are received using hawk-eye technique that uses six cameras. Achieving accurate results from the app each time was dependent on the background noise and the position of the camera and device quality.
- We informed user if the camera quality and device was not compatible for 240FPS and slow-mo recording was not available we did not allow them to download the app. A list of supported devices was also released on the Machineroad’s site.Â
- Ability to process videos recorded in different environments under different lighting conditions and pitches
- We did Machine learning for which we trained the model in different conditions like day and night, Outdoors and indoor etc. But it was difficult for the app to pick a new location and pitch automatically if it was different from the models on which it was trained.Â
- User needs to align the pitch with the camera and orientation of the camera should also be accurate else the results gets impacted.
- Help screens were implemented to inform user on the orientation and Video tutorials were released by the client as well so users can understand how to get the best results from the app.Â
Project Completion
Duration
Jan 2022 Beta release
Dec 2022 Actual release
Ongoing Support since Jan 2023
Deliverables
- Published the app on Play Store and Apple Store for beta users and then open it for all users and all regions.Â
- Deployment using AWS architecture on the cloud to setup a scalable and optimized backend system. It was configured to scale upto 6 X times the inital setup if more users start to use the app and if the Computing and storage resources are consumed fast enough. Â
- Integration with native camera capabilities with advance machine learning algorithm incorporated to get accurate speed and trajectories
- A test report along with On field demo with the client was done a few times to ensure the app works as expected. The average speed accuracy was set out to be 90% or more calculated for 20 bowls in the same location at the same time.Â
Support
As part of the project implementation we provided 1 month of extended support. This includes any Major / Minor bug fixes. And a further extended support for some issues where we have been supporting for a few years now.Â
Testimonial - AI ML based mobile app
We are awaiting a documented feedback from the clientÂ
Peritos have been have been very supportive of our business venture over the last 2-3 years. They have showing a willingness to invest their time in learning new technologies, which has helped us progress our application to where it is today. We look forward to continuing to work with them moving forward.
Mitch Ferguson
Co-Founder machineroad
Next Phase - AI ML based Mobile app
We are looking at doing the next phase of development and are already in POC stage whereÂ
- All the post processing of the video to be done on the mobile device itself this improves the chances for showing result faster to the end user.Â
- Implementing new features and rolling out new releases as part of our support agreement.Â