Brand Air

Project Title

Brand Air

Project Area of Specialization Artificial IntelligenceProject Summary

Brand Air is a platform which is an AI powered solution provider which will help the sponsors to add value to their investments with machine learning algorithm for robust image recognition using Computer vision that is an artificial intelligence field which deals with training computers to elucidate the content of images and video and perceive the visual world using deep learning models and can identify and classify objects accurately. From recognizing objects to processing the live action of any sport event, computer vision technology has been widely applied in many areas. Due to this advancement todays systems are generating results more quickly and accurately than humans; the accuracy rates of object detection and classification have gone from 50% to 99% in few years. Some of the algorithms for this fields are (RCNN, YOLO, SDD).

In this project we address the problem and hardships face by the Sponsors who invest their money in organizations. The main aim of the project is to solve their manual evaluation done by the workers in a way that it recognizes company’s ad through logo and audio detection in broadcast of sports event especially cricket match, it records the timing of on-air ad and generates report that will support business analysts in their task of sponsorship evaluation and further analysis according to the packages that user selects. The conclusion which is drawn is to provide a platform where calculation can be done efficiently by avoiding mistakes, inaccuracy, and miscalculations.

This application allows user to give input in the form that user can upload the video or can select the YouTube video for the detection of logo and audio purpose

This application will help the users to improve productivity of their investments operated by extracting data from images, live video streams and video frames.

Project Objectives

Feature of our project Brand Air are:

1. Logo Detection
2. Audio Detection
3. Advertisement Detection
4. Analysis report of advertisement

Solution that we addressed through this project are following: Brand air is a platform that will be able to

• Recognize company’s ad through logo and audio detection in broadcasts of sports event especially cricket match.
• To record the timing of on air – ad.
• Calculate the number of times the logo appeared.
• Generates a report that will support business analysts in their task of sponsorship. Evaluation and further analysis.

Project Implementation Method

Our project implementation is mainly focused on these three features:

LOGO DETECTION:
    Logo detection will be done by using YOLO  v3 which is an object detection algorithm.
Firstly frames will be extracted from the provided video and each frame will be sent as an input to the algorithm. The algorithm will  detect the logos from the frame and update the logo appearance count.
Figure provided below is an illustration of how YOLO works.

AUDIO DETECTION:
    Audio detection will be done by using the Google Speech API.
Firstly audio will be extracted from the video using python packages. Then the audio file will be sent as an input to the api that will return the text from the audio as a response. After extracting the text, user provided keywords will be matched from the text.

This project :

1. Will benefit the people who sponsors an event.
2. Will eliminate the time consuming efforts to analyze the on screen sponsorship perks.
3. Will be useful for those who can't afford paid evaluators.
4. Will benefit the user in terms of time and performance.
5. Will provide the report with error free results.

ADVERTISEMENT DETECTION:
    Advertisement detection will be done by comparing frames through grayscale comparison.
Firstly frames will be extracted from the video. Then each frame from the video will be compared with the user ad's frame and will count the matched frames through which the time duration of the ad will be calculated.

Benefits of the Project

Benefits of the project: 
This platform will automate the process of identifying essential context such as it monitors advertisements , detects logo  and the number of times it appeared in the form of analytical report  where promised perks can be calculated automatically without errors and mistakes as it is replacing manual evaluation. Less time will be required for the business growth . Money can be earned through this app as user will select packages. Customers trust can be gained easily as it will be easy in use and highly secured. Brand recognition will be gained which  is one of significant marketing goals. Sponsorship ROI will also be monitored.

Technical Details of Final Deliverable Technical Details of Final Deliverables: DELIVERABLES:

AUTHENTICATION MODULE:
    Users will register their account by using this module. Users will specify their company’s information along with unique email and password. The user will also select the package for its brand (silver, gold, platinum). This module will manage the authenticity of the user.

DATASET GATHERING MODULE:
    This module will allow the user to upload their brand’s branding material like logos or ads and will generate the data that will be used for model training. 

BRANDING MANAGEMENT MODULE:
    This module will detect logos, ads and keywords of the brand from the downloaded video as well as a YouTube video. This module will keep the record of logos and ads appearance and will keep a record of the brand name from the audio of the video with timestamps.

REPORT MANAGEMENT MODULE:
    This module will generate and manage the report from the attributes obtained from the Branding Management Module. This generated report will contain all the information extracted from the video.

PAYMENT MODULE:
    This module will integrate the payment gateway that will allow the user to pay for the packages that they have selected.

DEPLOYMENT DETAILS:

     This project will be a desktop application and will be integrated to the API developed on python that will be deployed on Cloud Server.

SUPPORT DOCUMENT:

    A complete report containing all the design and analysis of the system will be provided.

Final Deliverable of the Project Software SystemCore Industry MediaOther IndustriesCore Technology Artificial Intelligence(AI)Other TechnologiesSustainable Development Goals Partnerships to achieve the GoalRequired Resources