Human activity recognition for video surveillance using neural network /
Human activity recognition considered as one of the most effective technique in video surveillance due to it is the promising application in computer vision and signaling such as home care system, sign language, computer interaction and human machine. The main goal of video surveillance is to discov...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
Kuala Lumpur :
Kulliyyah of Engineering, International Islamic University Malaysia,
2017
|
| Subjects: | |
| Online Access: | http://studentrepo.iium.edu.my/handle/123456789/4858 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| LEADER | 036180000a22003010004500 | ||
|---|---|---|---|
| 008 | 180403s2017 my a f m 000 0 eng d | ||
| 040 | |a UIAM |b eng |e rda | ||
| 041 | |a eng | ||
| 043 | |a a-my--- | ||
| 050 | 0 | 0 | |a TK7882.P7 |
| 100 | 1 | |a Karm Allah, Mohanad Babiker Mohamed Osman, |e author | |
| 245 | 1 | 0 | |a Human activity recognition for video surveillance using neural network / |c by Mohanad Babiker Mohamed Osman Karm Allah |
| 264 | 1 | |a Kuala Lumpur : |b Kulliyyah of Engineering, International Islamic University Malaysia, |c 2017 | |
| 300 | |a xiii, 75 leaves : |b illustrations ; |c 30cm. | ||
| 336 | |2 rdacontent |a text | ||
| 347 | |2 rdaft |a text file |b PDF | ||
| 502 | |a Thesis (MSCE)--International Islamic University Malaysia, 2017. | ||
| 504 | |a Includes bibliographical references (leaves 68-71). | ||
| 520 | |a Human activity recognition considered as one of the most effective technique in video surveillance due to it is the promising application in computer vision and signaling such as home care system, sign language, computer interaction and human machine. The main goal of video surveillance is to discover the moving object and track their activities within the visible zone of the camera, however in the video surveillances human behavior analysis is a complex task because of their constantly various appearances, human crowds in different clothes and environments. In this research, we produce a novel method for human activity recognition were developed. The designed system has four stages, collect the data, construct the neural network, training, and testing. Three scenarios used in this research. Scenario one built based on a simple and clear background in an indoor environment to recognize the activity of walking, sitting, boxing and hand waving. The recognition of these activities was analyzed based on the basic features of the bounding box. Scenario two is more complex because of using very similar activities with challenging background which is provided in KTH dataset video. The features extracted by using spatial temporal interest point (STIP), centroid, and bounding box. The third scenario assumed to ensure the reliability of the used features in scenario two. Finally, the result shows that the assumption of scenario one highly success to recognize the activity of human the overall recognition rate is 98% .moreover, the result proves that in order to design an efficient recognition system via a neural network the combination of STIP features with basic blob analysis features is recommended. The designed system has achieved accuracy 90% on the six similar activities. In scenario two and scenario three, the recognition rate was 89.9% on nine different activities. At the end, the features of STIP are preferable to recognize the human activity than the basic feathers of Blob analysis because of it is the capability to distinguish between the similar activities in an accurate manner. | ||
| 596 | |a 1 | ||
| 655 | 7 | |a Theses, IIUM local | |
| 690 | |a Dissertations, Academic |x Department of Electrical and Computer Engineering |z IIUM | ||
| 710 | 2 | |a International Islamic University Malaysia. |b Department of Electrical and Computer Engineering | |
| 856 | 4 | |u http://studentrepo.iium.edu.my/handle/123456789/4858 | |
| 900 | |a sbh-aaz-ls | ||
| 999 | |c 440339 |d 472254 | ||
| 952 | |0 0 |6 T TK 007882 P7 K18H 2017 |7 0 |8 THESES |9 764138 |a IIUM |b IIUM |c MULTIMEDIA |g 0.00 |o t TK 7882 P7 K18H 2017 |p 11100379943 |r 2018-10-02 |t 1 |v 0.00 |y THESIS | ||
| 952 | |0 0 |6 TS CDF TK 7882 P7 K18H 2017 |7 0 |8 THESES |9 856650 |a IIUM |b IIUM |c MULTIMEDIA |g 0.00 |o ts cdf TK 7882 P7 K18H 2017 |p 11100379944 |r 2018-10-02 |t 1 |v 0.00 |y THESISDIG | ||