The document is a report on thesis study progression submitted by Shahriar Jahan and Roknuzzaman to their assistant professor. It discusses their work using the UCF101 video dataset for their thesis on vision-based human activity recognition with machine learning algorithms. They are preparing the dataset by converting each video in the UCF101 set to sequential image files, with the frame rate determining the number of images generated per second of video. They are currently working on the dataset preparation.
The document is a report on thesis study progression submitted by Shahriar Jahan and Roknuzzaman to their assistant professor. It discusses their work using the UCF101 video dataset for their thesis on vision-based human activity recognition with machine learning algorithms. They are preparing the dataset by converting each video in the UCF101 set to sequential image files, with the frame rate determining the number of images generated per second of video. They are currently working on the dataset preparation.
The document is a report on thesis study progression submitted by Shahriar Jahan and Roknuzzaman to their assistant professor. It discusses their work using the UCF101 video dataset for their thesis on vision-based human activity recognition with machine learning algorithms. They are preparing the dataset by converting each video in the UCF101 set to sequential image files, with the frame rate determining the number of images generated per second of video. They are currently working on the dataset preparation.
Title: Vision based human activity recognition using machine learning algorithm.
Course No: MTE 4100
Course Name: Project and Thesis
Submitted By Submitted To
Shahriar Jahan MD. Robiul Islam
Roll: 1608003 Assistant Professor Roknuzzaman Department of Mechatronics Engineering Roll: 1608005 RUET. Dataset preparation: The selected Dataset for our thesis is the UCF101 video dataset. It consists of 101 video classes with a total number of 13320 videos. All the videos are in ‘.avi’ file format. For our thesis work we need to convert each video file into sequential image files. The number of frame rate (fps) would determine the number of images that would generate form each second of the video. For example, if a video has 15fps frame rate, we would get 15 images from every second of that video and if the video is of 10 second time duration, then we would get a number of total 150 images. We are currently working on it.