Authors

Presenter(s)

Sankarshan Dasgupta

Files

Download

Download Project (332 KB)

Description

We have technologies implemented which assist us beyond the reasonable measures to make our life progress with minimum efforts and maximum output. Only lagging in our fight against devastating natural processes of Earth. The unstoppable force is beyond human intervention or control. Keeping in mind the disastrous effects of natural calamities, we propose an idea to help and assist the heroes of our real life such as Fire fighter, Defense personnel, Coast Guard etc. to save human lives, minimizing their individual risk. Detecting human in the distracted environment is very challenging due to the occlusion (i.e. people may be under debris), the unclear boundary (i.e. noisy background), and the coarse scale due to the distance. Here, we are building a network model and training it to seamlessly detect human out in plain sight. I am also considering this topic as part of my master thesis advised by Dr. Van Tam Nguyen. Creating a new data-set and then training the model to work precisely in harsh kind of environment possible. Images for data-set are chosen for their traits and uniqueness. Then feeding through a network built on Tensor-flow to learn from the data-set. Implementation of algorithm, since the data-set has new structure, shape and new observation of human images, algorithm chosen needs to be an accurate process with edge detection and processing. Then using it to up Mask R-CNN performance. We evaluate the proposed framework on the newly collected data-set. The extensive experiments on the data-set will evaluate the effectiveness of our proposed framework for this interesting problem. Unlike other object detection where the applicable methods of detection has significant references, human detection has been without an availability of well-formed data-set, which could set a benchmark and open the scope for many new research possibilities and high accuracy performing networks.

Publication Date

4-22-2020

Project Designation

Graduate Research

Primary Advisor

Van Tam Nguyen

Primary Advisor's Department

Computer Science

Keywords

Stander Symposium project, College of Arts and Sciences

United Nations Sustainable Development Goals

Quality Education; Industry, Innovation, and Infrastructure

Finding Human in Challenging Environment

Share

COinS