Magnifi Documentation

Feature Releases

Master Document: Knowledge Base

API Docs

Magnifi API Document

Folder for New Pages

Explore

Magnifi knowledge Base

AI Documentation

Last edited 485 days ago by Vinay Reddy

Introduction

This documentation provides an overview of the AI models used for various events and verticals, including Computer Vision (CV), Natural Language Processing (NLP), and Optical Character Recognition (OCR). Additionally, it covers the batsman celebration detection for 50/100 runs and the concept of start frame and end frame.

Events and Verticals

The AI system incorporates models for analyzing the following events and verticals.

4, 6, and Wickets

For detecting events like 4s, 6s, and wickets, the system employs a combination of computer vision techniques. It uses image or video analysis algorithms to identify specific actions such as boundary shots (4s and 6s) and wicket-taking moments.

CV (Computer Vision)

The CV component of the AI system is responsible for processing and analyzing visual data. It utilizes deep learning models such as convolutional neural networks (CNNs) to extract features and classify various objects, actions, or events in images or videos.

NLP (Natural Language Processing)

The NLP component of the AI system focuses on understanding and processing human language. It employs techniques like text classification, sentiment analysis, named entity recognition, and language modeling to extract meaningful information from textual data.

OCR (Optical Character Recognition)

OCR is used to recognize and extract text from images or scanned documents. The AI system employs OCR models to convert text-containing images into machine-readable text, enabling further analysis and processing.

Batsman Celebration for 50/100 Runs

The AI system is designed to detect and analyze batsman celebrations specifically for scoring 50 or 100 runs. It uses computer vision algorithms to recognize the unique patterns of celebratory actions associated with reaching these milestones. By analyzing visual cues, the system can identify and flag such celebrations during a match.

Start Frame and End Frame

The concept of start frame and end frame refers to the specific frames or timestamps within a video sequence that mark the beginning and end of a particular event or action. For example, in the context of boundary shots (4s and 6s) or wickets, the start frame would represent the frame where the action starts (e.g., the ball leaving the bat), and the end frame would indicate the frame where the action concludes (e.g., the ball crosses the boundary or wicket is taken).

The AI system leverages these start and end frames to precisely identify and analyze events of interest. By utilizing frame-level analysis, the system can provide accurate and granular insights into specific moments during a match.

AI Process Workflow

This workflow provides a concise overview of the processes involved in enhancing the work and addressing issues during the Proof of Concept (POC) stage. By understanding these processes, users can actively contribute to improving the work and effectively handle any challenges that may arise.

Enhancement Process

The enhancement process aims to increase or improve the quality and value of the work. It involves various steps and inputs, including

Step 1: Existing Client Enhancement

When it comes to enhancing the work, the following steps are involved:

Input Feed: Engineers with good knowledge of the work provide input feed, which includes video solutions and cloud engineering.

Input Metadata: Video solutions engineers provide additional metadata to enrich the input data.

Post Editing: Both video solutions and cloud engineers contribute to the post-editing phase.

Step 2: Internal Enhancement

Infrastructure Scaling: Scaling the infrastructure to accommodate increased workload or demands.

Optimization: Identifying and implementing the best practices and approaches.

Research and Development (R&D): Exploring new techniques and technologies to enhance the work.

Risk Enhancement: Mitigating risks and improving the overall reliability and security.

Step 3: QA Enhancement

Quality Assurance processes are implemented to ensure the enhancements meet the desired standards and requirements.

Issues in POC Process

The process addresses issues encountered during Proof of Concept (POC) and post-POC stages. It involves the following steps:

During POC and Post POC

Input Feed: Video solutions engineers and cloud engineers provide the necessary data or information as input to the AI system or model being developed or tested.

Input Metadata: Video solutions engineers enrich the input data by providing supporting information and additional insights.

Post Clip Work: Various tasks such as generating highlights, editing single clips, publishing, and creating web stories are performed.

Enhancements: Timelines are fixed, and work is done to address any necessary improvements or fixes.

Post POC

Logged Issues: Identified issues during the POC and post-POC stages are logged to analyze the reasons behind their occurrence.

Not Logged Issues: Any issues that were not previously logged are identified and logged for further investigation and resolution.

Team Composition

The team consists of members from different sports disciplines, including basketball, tennis, football, and cricket. Each team is responsible for specific areas related to video and cloud solutions, application development, machine learning (ML), and research and development (R&D).

Here is the breakdown of the teammates across the respective sports teams:

Basketball Team

Team Leader: Mayank

Members: Vinay, Ayush Patidar, Kushal

Tennis Team

Team Leader: Jay

Members: Ajeet, Sai

Football Team

Team Leader: Vivek

Members: Prabhat, Pravin, Omkar

Cricket Team

Team Leader: Jay

Members: Prabhat, Alankar, Manisharan , Shyam

Want to print your doc?
This is not the way.

Try clicking the ⋯ next to your doc name or using a keyboard shortcut (

CtrlP

) instead.