Online object detection and tracking
The report introduces the real-time object detection and tracking by using HOG feature extraction, mean-shift tracking and k-NN classification methods. Detail explanation on each method used will be done in the subsequent chapters. Furthermore, MatLab is used as programming platform for this trackin...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project (FYP) |
| Language: | English |
| Published: |
2013
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10356/54220 |
| _version_ | 1824455683740794880 |
|---|---|
| author | Ooi, Jun Han. |
| author2 | Teoh Eam Khwang |
| author_facet | Teoh Eam Khwang Ooi, Jun Han. |
| author_sort | Ooi, Jun Han. |
| collection | NTU |
| description | The report introduces the real-time object detection and tracking by using HOG feature extraction, mean-shift tracking and k-NN classification methods. Detail explanation on each method used will be done in the subsequent chapters. Furthermore, MatLab is used as programming platform for this tracking system.
The tracking system is specifically designed to detect the motion of vehicle in a video. There are 5 stages in the tracking system, determination of vehicle’s coordinate (done manually), database generation, mean-shift coordination, determination of accuracy and tracking error correction.
The tracking system is initialized by manually determine the X and Y coordinate of the car from the first frame of the video. By using HOG feature extraction algorithm, a database containing positive (vehicle’s features) and negative (background features) histograms (540 bins) is generated.
During the tracking stage, the vehicle’s coordinate is predicted using mean-shift algorithm. Further analysis and error corrections are done with k-NN classifier algorithm to increase the accuracy of the tracking system. The tracking process is then repeated until the last frame of the video.
Finally, the full process of the tracking system is observed and evaluated. The recommended solutions to improve the efficiency of the tracking system are also provided in the report for future research. |
| first_indexed | 2025-02-19T03:42:06Z |
| format | Final Year Project (FYP) |
| id | ntu-10356/54220 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2025-02-19T03:42:06Z |
| publishDate | 2013 |
| record_format | dspace |
| spelling | ntu-10356/542202023-07-07T17:32:05Z Online object detection and tracking Ooi, Jun Han. Teoh Eam Khwang School of Electrical and Electronic Engineering DRNTU::Engineering::Electrical and electronic engineering The report introduces the real-time object detection and tracking by using HOG feature extraction, mean-shift tracking and k-NN classification methods. Detail explanation on each method used will be done in the subsequent chapters. Furthermore, MatLab is used as programming platform for this tracking system. The tracking system is specifically designed to detect the motion of vehicle in a video. There are 5 stages in the tracking system, determination of vehicle’s coordinate (done manually), database generation, mean-shift coordination, determination of accuracy and tracking error correction. The tracking system is initialized by manually determine the X and Y coordinate of the car from the first frame of the video. By using HOG feature extraction algorithm, a database containing positive (vehicle’s features) and negative (background features) histograms (540 bins) is generated. During the tracking stage, the vehicle’s coordinate is predicted using mean-shift algorithm. Further analysis and error corrections are done with k-NN classifier algorithm to increase the accuracy of the tracking system. The tracking process is then repeated until the last frame of the video. Finally, the full process of the tracking system is observed and evaluated. The recommended solutions to improve the efficiency of the tracking system are also provided in the report for future research. Bachelor of Engineering 2013-06-17T03:56:55Z 2013-06-17T03:56:55Z 2013 2013 Final Year Project (FYP) http://hdl.handle.net/10356/54220 en Nanyang Technological University 134 p. application/pdf |
| spellingShingle | DRNTU::Engineering::Electrical and electronic engineering Ooi, Jun Han. Online object detection and tracking |
| title | Online object detection and tracking |
| title_full | Online object detection and tracking |
| title_fullStr | Online object detection and tracking |
| title_full_unstemmed | Online object detection and tracking |
| title_short | Online object detection and tracking |
| title_sort | online object detection and tracking |
| topic | DRNTU::Engineering::Electrical and electronic engineering |
| url | http://hdl.handle.net/10356/54220 |
| work_keys_str_mv | AT ooijunhan onlineobjectdetectionandtracking |