Now 7108 visitors
Today:0 Yesterday:0
Total: 156717
216S 51P 5R
2026-10-22, Week 43 |
| Member Login |
| TACT Journal Homepage |
| Call for Paper |
| Committee |
| Find My Paper |
| Author Homepage |
| Statistics & History |
| Welcome Message |
| Paper Procedure |
| FAQ |
| Paper Archives |
| Outstanding Papers |
| Author Homepage |
| - Paper Procedure |
| - Journal Procedure |
| - Presentation Tips |
|
| Program & Proceedings |
| Program with Papers |
| Plenary Session |
| Tutorial Session |
|
| Presentation Platform |
| Hotel & Travel Info |
| About Korea |
| Accommodation |
| Transportation |
| VISA |
| Other Infomation |
|
| Photo Gallery |
| Scheduler Login |
| Seminar |
| Archives Login |
|
| Sponsors |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Save Q&A |
 |
* Edit or answer any Q&A by selecting Q&A number Hyper Link below + Write button (Save)
|
ICACT20230170 Question.10
Questioner: m11007503@gapps.ntust.edu.tw 2023-02-18 오후 8:10:24 |
ICACT20230170 Answer.10
Answer by Auhor phamdinhlam@kyonggi.ac.kr 2023-02-18 오후 8:10:24
Chrome  Click!! |
| Thank you for the explanation, In the slide 10. The architecture uses real time frame processing with multiple CCTV source. By using high frame from the CCTV video also process the frame to detect object using YOLO. The question is how this system handle computational constraints that must be handle in real time while maintaining system performance to display CCTV video in real time? Thank you. |
Thank you for your great question.
As we know, keeping system performance to display CCTV video in real-time while using YOLO to detect object in the video is the difficult task. In our system, first of all we don't display all CCTV cameras. The system only detects object in the video using YOLO and stores these information in a JSV format to retrieval content from the video in the future. In case of displaying some CCTV camera from requesting of the security officer, the system will only display the requested cameras. Furthermore, all the processing videos are processed in a queue using different CPU threads.
Our system is also under updated for displaying more CCTV camera in real-time by using some video compressing techniques before detecting objects.
|
ICACT20230170 Question.2
Questioner: jianms@nfu.edu.tw 2023-02-20 오후 1:49:52 |
ICACT20230170 Answer.2
Answer by Auhor phamdinhlam@kyonggi.ac.kr 2023-02-20 오후 1:49:52
Chrome Click!! |
| Dear Sir,
Is there any evaluation about the performance of the object contentualizing to the video? How much time should be spent for the framing and object recognition?
Or how about the hardware resource needed for the performance?
|
Thank you for your question. In this study, we focused on proposing the architecture and methodology for explaining how to extract the detected information from the objects of the CCTV video in real time. Furthermore, we introduced and proposed the JSON video format or JSV for storing detailed information about the objects from video frames. This approach can be used with YOLO, SSD, or other object detection methods. YOLO is a case used in this study. Comparing the performance between different methods and how to choose the best model for training and predictions will be studied in the future research. |
ICACT20230170 Question.8
Questioner: sdk@etri.re.kr 2023-02-20 오후 4:30:13 |
ICACT20230170 Answer.8
Answer by Auhor phamdinhlam@kyonggi.ac.kr 2023-02-20 오후 4:30:13
Chrome Click!! |
| The role of the real-time frame processing component in this system seems to be to store it in the video store or send it to the object recognition component. It seems that the way video frames are handled is determined by priority of frame. Please let me know how the priority of frame is set. |
Thank you for your question. As you can see in the contextual schema of the JSV file on slide 11, all the information about a specific CCTV camera is stored in the metadata section. It includes but not limited to location, timestamps, FPS, resolutions, etc. The ID of a specific frame is stored in Frame -> Properties -> ID |
| |
