Poster Presentation [University of Dhaka]- Implementation Techniques of Incorporating face annotation scheme to help user search photo in personal photo frame application.
Download as PPT, PDF1 like813 views
This document discusses incorporating facial annotation into a personal photo frame application to help users search photos. The system allows users to manipulate and search photos based on facial appearance. It provides an image auto-annotation fusion scheme using existing annotation algorithms and face detection methods to efficiently annotate faces and estimate attributes. Users can manually annotate faces by drawing on the canvas and adding tags. The system was tested on 250 faces in 525 photos, achieving 80% correct automatic annotation and high precision for positional searches centered on faces.
![Incorporating face annotation scheme to help user search photo in personal photo frame application
Showrav Mazumder, Fuad Hossain, Dr. Supratip Ghose*
Department of CSE
University of Information Technology & Sciences
Introduction
In recent years, People use web space to keep the personal photos.
In this paper we have deployed a Photo Frame System (PFS) that allow the user to
manipulate and search photo based on their facial appearance conforming user
experience
The aim of this research is to help users maintain a personal window of photos based
on the image auto annotation with the help of the research of content based image
retrieval.
The contribution is that we provide a fusion scheme for efficient image auto
annotation and flexible manner estimating face attributes and face similarity.
System Overview
1. The proposed application based on the existing annotation algorithms in the
automatic and manual form.
2. Our system consists of Browse panel to browse the windows; Search Panel to
search photos and Mode Panel to manage people appeared in a photo in Manual and
auto-annotation mode.
3. Currently, the user can formulate their plan by drawing blobs onto the canvas
while tagging. In our system, we scaled off the picture by indexing it with the width
and the height and then tags it with <user, location> tag.
4. For users drag and scale system presented to change positions and the sizes of the
face and form search intention automatically. The user can bring the canvas to
specify the regions of face appearances and add a tag for the person in the registered
list in contact API.
Conclusion
Our system is simple and easy not only to maintain personal memories by also to
search photos.
By drawing the position with tags, we seamlessly connect spatial information with
people in photos and enable more flexible photo lookup, according to the user‘s
intention.
Result Overview
1. Toward the end, in this paper, for the performance evaluation we have kept 250
faces in 525 daily photos in the dataset.
2. For a given semantic descriptor, positional query corresponding to the center
returns the highest precision .70 whereas the top left and top right returns the
accuracy of around .63 and .62. In image auto-annotation, the total correct
annotation rate is 80% making the fusion scheme useful.
3. We are currently planning more aesthetic aspect of the image and employing
the ranking algorithm in image automation. Two correct annotation result given
in below:
Figure 2. Result of Correct Annotation. Figure 3. Another Result of Correct Annotation.
Related Works
1.Image annotation is done by a fusion scheme of Cascaded HAAR classifier [1] and
color based facial region likelihood classifier [2] to reduce false positive.
2.Fusion scheme with the ANN classifier serves a good filter in the geometric
component Adboost HAAR classifier.
3.In this way we overcome the problems of manual text based search [3].
Figure 1. PFS front end for images annotation.
References
[1] Viola, P., and Jones, M.,―”Rapid object detection using boosted cascade of simple features,”
IEEE Conference on Computer Vision and Pattern Recognition, in Proc. CVPR’01, vol. 1, pp. I-
511, 2001.
[2] Wilson, P. I., Fernandez, and J. (n. d.), ―”Facial feature detection using Haar classifiers,” Journal
of Computing Sciences in Colleges 21, no. 4, pp. 127–133, 2006.
[3] T. Sumathi, C. Lakshmi Devasena, and M. Hemalatha, “An overview of automated image
annotation approaches,” International Journal of Research and Reviews in information
sciences, Science Academy Publisher, vol. 1, no. 1, pp. 231-245, 2011.](https://image.slidesharecdn.com/postertemplatedu-150923100733-lva1-app6892/85/Poster-Presentation-University-of-Dhaka-Implementation-Techniques-of-Incorporating-face-annotation-scheme-to-help-user-search-photo-in-personal-photo-frame-application-1-320.jpg)
Poster Presentation [University of Dhaka]- Implementation Techniques of Incorporating face annotation scheme to help user search photo in personal photo frame application.
- 1. Incorporating face annotation scheme to help user search photo in personal photo frame application Showrav Mazumder, Fuad Hossain, Dr. Supratip Ghose* Department of CSE University of Information Technology & Sciences Introduction In recent years, People use web space to keep the personal photos. In this paper we have deployed a Photo Frame System (PFS) that allow the user to manipulate and search photo based on their facial appearance conforming user experience The aim of this research is to help users maintain a personal window of photos based on the image auto annotation with the help of the research of content based image retrieval. The contribution is that we provide a fusion scheme for efficient image auto annotation and flexible manner estimating face attributes and face similarity. System Overview 1. The proposed application based on the existing annotation algorithms in the automatic and manual form. 2. Our system consists of Browse panel to browse the windows; Search Panel to search photos and Mode Panel to manage people appeared in a photo in Manual and auto-annotation mode. 3. Currently, the user can formulate their plan by drawing blobs onto the canvas while tagging. In our system, we scaled off the picture by indexing it with the width and the height and then tags it with <user, location> tag. 4. For users drag and scale system presented to change positions and the sizes of the face and form search intention automatically. The user can bring the canvas to specify the regions of face appearances and add a tag for the person in the registered list in contact API. Conclusion Our system is simple and easy not only to maintain personal memories by also to search photos. By drawing the position with tags, we seamlessly connect spatial information with people in photos and enable more flexible photo lookup, according to the user‘s intention. Result Overview 1. Toward the end, in this paper, for the performance evaluation we have kept 250 faces in 525 daily photos in the dataset. 2. For a given semantic descriptor, positional query corresponding to the center returns the highest precision .70 whereas the top left and top right returns the accuracy of around .63 and .62. In image auto-annotation, the total correct annotation rate is 80% making the fusion scheme useful. 3. We are currently planning more aesthetic aspect of the image and employing the ranking algorithm in image automation. Two correct annotation result given in below: Figure 2. Result of Correct Annotation. Figure 3. Another Result of Correct Annotation. Related Works 1.Image annotation is done by a fusion scheme of Cascaded HAAR classifier [1] and color based facial region likelihood classifier [2] to reduce false positive. 2.Fusion scheme with the ANN classifier serves a good filter in the geometric component Adboost HAAR classifier. 3.In this way we overcome the problems of manual text based search [3]. Figure 1. PFS front end for images annotation. References [1] Viola, P., and Jones, M.,―”Rapid object detection using boosted cascade of simple features,” IEEE Conference on Computer Vision and Pattern Recognition, in Proc. CVPR’01, vol. 1, pp. I- 511, 2001. [2] Wilson, P. I., Fernandez, and J. (n. d.), ―”Facial feature detection using Haar classifiers,” Journal of Computing Sciences in Colleges 21, no. 4, pp. 127–133, 2006. [3] T. Sumathi, C. Lakshmi Devasena, and M. Hemalatha, “An overview of automated image annotation approaches,” International Journal of Research and Reviews in information sciences, Science Academy Publisher, vol. 1, no. 1, pp. 231-245, 2011.