AI edge device based human action recognition for mobile platforms
Deep learning has contributed to a huge improvement to the ever-developing field of Computer Vision. Many state-of-the-art applications like face recognition or machine vision in self-driving cars have been introduced based on Computer Vision. In real world cases, we find that an activity does not o...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
Nanyang Technological University
2021
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/150048 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | Deep learning has contributed to a huge improvement to the ever-developing field of Computer Vision. Many state-of-the-art applications like face recognition or machine vision in self-driving cars have been introduced based on Computer Vision. In real world cases, we find that an activity does not occur alone by itself but as an interaction between people and objects. Humans are able to distinguish such actions as we are able to associate them with our ability to apply logic and reason. However, a computer does not have such association capabilities. With the fast advancements in Computer Vision, there are also discussions about privacy concerns due to the large datasets necessary for deep learning. These datasets may contain sensitive information which is prone to vulnerabilities when processed on the cloud. In this paper, we design an activity recognition model by combining an action recognition and object detection model together and implement it on an edge device to tackle these issues. We will also discuss on the key findings and the challenges and future work that could be done. |
|---|