3. Strength and Conditioning: Elevating Athletic Performance
Our strength and conditioning program blends cutting-edge science with personalised training regimes to optimise athletic potential. We explore fully non-invasive computer vision based approaches to examine various strength and conditioning exercises ranging from squats to weightlifting.

4. Smart Insole System With IoT For Performance Monitoring In Cricket And Boxing.
Smart Insole is an IoT and cloud based online gait analysis platform that provides quantitative analytics using metrics along with correlated video. key metrics include plantar pressure, COP, weight distribution from cost-effective force sensitive resistor sensors (FSR) and stride, swing, stance variabilities, asymmetries and foot angle from inertial measurement units (IMU). The developed platform can be used extensively in sports and medical domains for assessing player performances or patient recovery and rehab.

Analytics: Smart insole application offers real-time and post analytics. Realtime provides instantaneous pressure points as heat maps, COP trajectory, step count, stride length and velocity, foot angle etc. Post analytics incorporates the above said parameters from an offline data with sever uploaded correlated video. Data transfer with higher throughput, analytics with more domain specific metrics, capacitive based pressure sensor arrays are some of the extended functionalities which are being worked on.

Current/Planned Use Case:

• Smart insole is currently used by Royal Challengers Bangalore for player performance and improvements and by MIOT hospital for patient recovery and rehab.
• Employing smart insole in different sports domains like tennis, badminton, boxing, wrestling and in fatigue management like DMS are some of the main planned use cases.

5. Performance Modelling For Archery.

6. AI-Enabled Video-Analysis System For Elite Athlete Performance Enhancement In Throwing Sports Like Javelin And Shot-Put.

7. SwimSense: Real-Time Bio-mechanical Feedback Using Computer Vision In Swimming.

8. Ball Tracking in 3D and Spin Recovery.

Updates:

• Hawk eye which is extensively employed in cricket performs real-time ball tracking in 3D. Our work is aimed at ball spin estimation in addition to tracking in 3D. We show that using two static cameras along with ball colour as a prior, we can estimate the 3D ball trajectory and spin (magnitude and axis of rotation) with only one distinguishable feature on the ball. We validate our method with experiments on cricket ball and football in indoor as well as outdoor environments. This work is under review in a journal.
• Temporal action localisation in boxing videos: The problem of temporal localisation in videos is of great interest to the sports fraternity. We are currently working on parsing boxing actions in videos using weak supervision from video- level labels alone. This is being posed as a multi-instance multi-label classification problem. Our initial efforts are focused on parsing at 1-sec intervals which we will then extend to frame-level parsing.This is ongoing work and has great potential in enabling search and retrieval of specific actions from long video sequences.

9. Dynamic Novel View Synthesis.
Going forward, we are interested in synthesising novel views for boxing, javelin and table tennis games using dynamicNeural Radiance Fields(NeRFs). Given  a video taken from a handheld camera, the goal is to generate videos from an arbitrary view point and for a specified time stamp. This is ongoing work.

10. Fan engagement.
Another area that we are investigating is is to use computer vision technology, specifically Augmented reality (AR) and Virtualreality (VR), to deliver an immersive fan experience. The idea is to make spectators feel like they are part of the game. Features such as panoramic camera angles, real-time statistics, and on-demand replays are now possible. As ongoing work, we are exploring other capabilities to enhancefan engagement.