Deep learning with sparse labels: Activity detection is an important problem which has a wide range of applications in surveillance, security, crowd management, etc. The exising methods based on deep learning requires a large amount of annotated training samples to develop an activity detection method. The goal of the project is to develop deep learning method to perform activity detection in videos where we want to use few annotations. This will also involve techniques from active learning. The student will learn how active learning can be used to gather data annotations in an effective way where instead of annotating all the data samples, we assign labels to some of the representative samples.
Project Dates
Start Date: 1/11/2023 - End Date: 4/27/2023
Students Needed
Type of Project
Individual
Student Responsibilities
The student will start with basics of deep learning and computer vision. The student will later be responsible to perform some experiments which will be designed for performing action detection in videos using sparse labels.
Time Commitment
10 hours hour(s)
Student Requirements
Good programming skills, Self motivated, Organized, Interest in Computer Vision and Deep learning, The student should have a good understanding of Python coding. It will be highly used during this project.
Interested in Working With the Following Programs
For EXCEL URE Students Only
Additional Notes