AI based event coverage

In our daily routine, we have seen that for any sort of event coverage, we need several labour. So it makes very uneasy for the guests and makes it hard to manage for the event manager, for instance as we have one person for video recording and for capturing pictures we have another person, and some

Project Title

AI based event coverage

Project Area of Specialization

Artificial Intelligence

Project Summary

In our daily routine, we have seen that for any sort of event coverage, we need several labour. So it makes very uneasy for the guests and makes it hard to manage for the event manager, for instance as we have one person for video recording and for capturing pictures we have another person, and some of the times a same person is assigned all of the tasks all by himself and he have to all of them serially which requires a lot of time and tiring work. So the idea was based on the thought that, “Why shouldn’t we make this coverage autonomous by using a drone and let it bear all the hassle”. To perform these tasks and make the drone autonomous, we have used Hand Gestures recognition. Drone will select and operate on different modes based on the gesture provided to it by the user.The drone after selecting a particular mode will performs different tasks that we have assigned, e.g. face recognition and tracking, surveillance, beacon, and object detection, obstacle avoidance, live videos coverage and capturing pictures. That is how it will help the people to reduce the personnel required and move forward with new technologies.

Project Objectives

The objectives of this project are: 

• To collect the data for object detection and face Recognition for machine learning. 
• To preprocess and make the drone recognize the object using a machine learning/deep learning model. 
• To preprocess and make the drone recognize and track the face using a machine learning/deep learning model.
• To build the model for hand gestures recognition.
• To make sure the drone stabilizes on coordinates when recording. 
• To broadcast data preprocessed by machine learning model. To capture pictures and videos by an automated python script. 
• To reduce the personnel engaged in event coverage.

Project Implementation Method

We intend to expand a Software-based Application with a purpose to automate the drone. We will provide Hand Gestures to the drone, and it's going to act in step with what we've described withinside the capabilities like Object Detection, Drone Surveillance, Obstacle avoidance, Beacon Detection, and Face Recognition.
The drone features a distinguished feature that's Hand Gesture Recognition, and Here 1st we've got to create a model for Hand Gesture Recognition, and we have a tendency to have outlined some gestures in it, so the drone can perform per the gestures that we have defined in it, For this, we have used MediaPipe a library of python. The MediaPipe is a framework designed to implement production-ready machine learning that has got to build pipelines to perform logical thinking over impulsive sensory data, has printed code concomitant analysis work, and build technology prototypes.
For object detection we need a dataset by which we can detect objects, so we used a “Coco dataset”, this is a large scale object detection dataset. The method we are using is ‘Single Shot MultiBox Detector’ and ‘MobileNetV3’.
The main purpose of object detection is to identify and locate one or more effective targets from still images or video data. It comprehensively includes a variety of important techniques, such as image processing, pattern recognition, artificial intelligence and machine learning, For this, we have used the OpenCV library. It is a library of programming functions mainly used for image processing. We can solve many real-time problems using image processing applications, so by using OpenCV we are detecting multiple objects in real-time. We have created a function like that after we show the index of the hand which mode is to be selected. We are able to  detect object within the surveillance mode which is able to later facilitate us for obstacle avoidance in surveillance mode. For the Face Detection, the drone is ready to recognise the face of someone within the frame and follow or track it. In this mode, the drone focuses totally on the person or the other particular object. To perform this task, we have a tendency to use haar cascade algorithmic program that helps the drone to recognise a person or any other object, Facilitating the drone to maneuver in conjunction with the person and track their movement with none collision; a selected distance is to be obtained by the drone to induce a transparent and evident sight.

Benefits of the Project

The benefits for this project is that it helps the person doing coverage. The person has to take his cameraman everywhere to record the event or take a live coverage of the event, and sometimes it is complicated to arrange a person, which would result in time consumption, but using drones would surely be beneficial in many cases. It can help the media person that they are not devoid of any person. Using modern technology in drones, we would provide all such features that would not require any help while recording or taking live coverage. Drone, as compared to ground vehicle, can reach its destination in minimum time as it is free from obstruction caused by traffic or any other physical means as a drone can quickly go over/under an obstacle by predicting the shortest route to a destination waypoint. For extensive coverage areas, drones via bird's-eye view and some cool features can help the media person take a coverage independently without the need of any help.

Technical Details of Final Deliverable

The final deliverable is "Making drone smart to position itself by analyzing positioning parameters". Which means that once drone has taken off and assigned a particular mode by gesture. So it should evaluate the distance it has to perform coverage on, in both axes. This will be achieved by using beacons present in all four directions. Drone will detect and recognize them and calculate the distance between them using computer vision and start moving itself accordingly. The the final outcome of the drone will be in the form of following modes with relevant to the strategy defined above.

MODES:

1. Tracking face while maintaing it's position.
2. Autonomous event coverage mode while avoiding obstacles.
3. Surveillance mode ( detecting numerous objects in the environment while hovering).
4.  Capturing photographs by recognizing some posture/sentiment .

Final Deliverable of the Project

HW/SW integrated system

Core Industry

IT

Other Industries

Media

Core Technology

Artificial Intelligence(AI)

Other Technologies

Sustainable Development Goals

Industry, Innovation and Infrastructure

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com