Search & Rescue UAV Project

Objective

Develop a UAV-based system for search and rescue operations using computer vision, navigation, and autonomous control.

Description

Background

Autonomous image segmentation and feature detection and classification is an important aspect of image processing. Image segmentation is the process of “partitioning a digital image into multiple segments''. The goal of segmentation is to simplify and change the representation of an image into something that is more meaningful and easier to analyze. Feature detection includes finding areas of interests such as edges, corners and simple shapes. These features are then classified into various categories based on their shape, colour or other inherent features. These concepts are widely used in military and civilian UAV missions to gather information about areas out of human reach, such as disaster-stricken or mountainous areas.

Task

The theme for this task is Search and Rescue. A shipwreck has occurred in the ocean, and your job is to gather information about the location and condition of stranded passengers. Your UAV is collecting aerial images of the wreckage that look like the sample image given below. Information about the input image : ● The input image is divided into two primary regions: the blue region corresponds to ocean, while the brown/green region corresponds to land. Passengers within the image are denoted by geometric shapes, with stars representing children, squares representing adults, and triangles representing elderly individuals. ● The severity of each civilian’s condition is represented through color coding: red indicates severe condition, yellow indicates mild condition, and green indicates safe. ● Additionally, the image contains three designated rescue pads (zones for evacuation and safety denoted by circle). Among these, two rescue pads are situated on land and one rescue pad is located in water. While all rescue pads serve the same purpose, civilians are required to be assigned to the best available rescue pad based on their position in the image and their medical emergency while keeping in check the capacity.

Sample Image The task for you is to devise a method to assign each of the casualty to the best possible rescue camp while making sure that the final casualty configuration for each of the camp is their respective best possible combination based on the casualty scores. . This must be based on the following rules : Priority order of casualties : Star-3(Highest), Triangle-2, Square-1(Lowest) Priority order of emergency : Severe-3(Highest), Mild-2, Safe-1(Lowest) Max capacity of rescue camps : Pink-3 casualties, Blue-4 casualties, Grey-2 casualties The best rescue camp for a particular casualty is based on a final score calculated by the amalgamation of the Priority score and distance where the priority score is Priority(casualty*emergency). In case of a similar priority score , a casualty with higher emergency score will be given importance. Devise your own score taking in all the considerations and keep in mind the max capacity of the camps while making sure that each camp has the highest possible total priority score. Input A list of 10 images, similar to the sample image provided above Expected Output

1. An output image, for each input image, that clearly shows the difference between the ocean and land, by overlaying 2 unique colors on top of each. The expected output for the given sample input is given below.
2. a)Count the number of casualties assigned to each of the three camps. b)The details of casualties assigned to each of the three camps for each image(Agegroup ,medical emergency ) in the order [blue,pink,grey].
3. The total priority of each of the camps saved in a list and the avg. priority of the image (rescue ratio of priority Pr) , calculated by summing the priorities of the camps and averaging over the number of casualties.
4. A list of the names of the input images , arranges in descending order of their rescue ratio (Pr) The expected output for the given sample image is given below
5. Sample output (Arrows are used for explanation)
6. [ [[3,3],[1,2]] , [[2,2],[3,1]] , [[3,2],[2,2]] ] green star was assigned to pink camp due to lower priority than yellow triangle although distance must also be considered which will be based on your score formula.
7. [[2,2]] (Priority score of casualty = 2x2 = 4 (for safe elderly) and along with distance calculate your own defined score.
8. [[Summation of scores for blue] ,[Summation of scores for pink] ,[Summation of scores for grey] ]
9. image1, image3, image4….. etc (not related to the given sample image) Example: Input images:
   • Image1

• Image2

Sample Output: Segmented Images for ocean and land. Image_n= [ [[1,1],[1,2],[1,3],[3,2]] , [[3,1],[2,3],[3,3]] , [[2,1],[2,2]] ] for max 9 casualties. Camp_priority = [[24,45,56],[25,50,70]] Priority_ratio = [125/8=15.625,145/9=16.1] image_by_rescue _ratio = [Image2, Image1] To simplify the given task, we’ve given a step by step approach to learn various concepts and libraries that are required to complete the task.

Progress Tracking

This repository follows continuous progress commits. Daily progress reports are maintained in /reports.

Repository Structure

• src/ → Core modules
• reports/ → Progress reports
• docs/ → Diagrams, references and images

Mentor

Shivam Sharma

Built With

• python