Mining weakly labeled web facial images for search based face annotation
2 likes540 views
This paper presents a search-based face annotation framework that utilizes weakly labeled facial images from the web, addressing challenges in effective annotation through an unsupervised label refinement (ULR) approach. The proposed system improves upon existing methods by refining label quality and employing a clustering-based approximation to enhance scalability. Experimental results demonstrate the efficacy of the ULR technique, which outperforms traditional approaches, indicating potential for future research in supervised learning techniques.
Mining weakly labeled web facial images for search based face annotation
- 1. #13/19,1st Floor, Municipal Colony, Kangeyanellore Road, Gandhi Nagar, Vellore- 632006 website: shakastech.com, Gmail: Shakastech@gmail.com Phone No: 0416-6066663/2247353 Mobile No: 9500218218 MINING WEAKLY LABELED WEB FACIAL IMAGES FOR SEARCH-BASED FACE ANNOTATION ABSTRACT This paper investigates a framework of search-based face annotation by mining weakly labeled facial images that are freely available on the World Wide Web (WWW). One challenging problem for search-based face annotation scheme is how to effectively perform annotation by exploiting the list of most similar facial images and their weak labels that are often noisy and incomplete. To tackle this problem, we propose an effective unsupervised label refinement (ULR) approach for refining the labels of web facial images using machine learning techniques. We formulate the learning problem as a convex optimization and develop effective optimization algorithms to solve the large- scale learning task efficiently. To further speed up the proposed scheme, we also propose a clustering-based approximation algorithm which can improve the scalability considerably. EXISTING SYSTEM Existing object recognition techniques to train classification models from human- labeled training images or attempt to infer the correlation/probabilities between images and annotated keywords. Given limited training data, semi-supervised learning methods have also been used for image annotation Disadvantage of existing system 1. Local binary system not find clear image. 2. Its take lot of time for find the image PROPOSED SYSTEM We investigate and implement a promising search based face annotation scheme by mining large amount of weakly labeled facial images freely available on the WWW. . We propose a novel ULR scheme for enhancinglabel quality via a graph-based and low-rank learning approach.
- 2. #13/19,1st Floor, Municipal Colony, Kangeyanellore Road, Gandhi Nagar, Vellore- 632006 website: shakastech.com, Gmail: Shakastech@gmail.com Phone No: 0416-6066663/2247353 Mobile No: 9500218218 Advantage proposed system 1. Easily get the images using face code word from database. 2. Very faster than old system MODULE DESCRIPTION: 1. content-based image search 2. Face annotation 3. Face annotation performance on database 1. content-based image search: Content-based image retrieval (CBIR), also known as query by image content (QBIC) and content-based visual information retrieval (CBVIR) is the application of computer vision techniques to the image retrieval problem, that is, the problem of searching for digital images in large databases. 2. Face annotation The classical image annotation approaches usually apply some existing object recognition techniques to train classification models from human-labeled training images or attempt to infer the correlation/probabilities between images and annotated keywords. 3. Face annotation performance on database This experiment aims to verify the annotation performance of the proposed SBFA framework over a larger retrieval database: “DB1000.” As the test database is unchanged, the extra facial images in the retrieval database are definitely harmful to the nearest facial retrieval result for each query image. A similar result could also been observed where the mean average precision became smaller for a larger retrieval database.
- 3. #13/19,1st Floor, Municipal Colony, Kangeyanellore Road, Gandhi Nagar, Vellore- 632006 website: shakastech.com, Gmail: Shakastech@gmail.com Phone No: 0416-6066663/2247353 Mobile No: 9500218218 Architecture System Configuration:- H/W System Configuration:- Processor - Pentium –III Speed - 1.1 Ghz RAM - 256 MB(min) Hard Disk - 20 GB Floppy Drive - 1.44 MB Key Board - Standard Windows Keyboard Mouse - Two or Three Button Mouse Monitor - SVGA
- 4. #13/19,1st Floor, Municipal Colony, Kangeyanellore Road, Gandhi Nagar, Vellore- 632006 website: shakastech.com, Gmail: Shakastech@gmail.com Phone No: 0416-6066663/2247353 Mobile No: 9500218218 S/W System Configuration:- Operating System :Windows95/98/2000/XP Application Server : Tomcat5.0/6.X Front End : HTML, Java, Jsp Scripts : JavaScript. Server side Script : Java Server Pages. Database : Mysql Database Connectivity : JDBC. CONCLUSION This paper investigated a promising search-based face annotation framework, in which we focused on tackling the critical problem of enhancing the label quality and proposed a ULR algorithm. To further improve the scalability, we also proposed a clustering-based approximation solution, which successfully accelerated the optimization task without introducing much performance degradation. From an extensive set of experiments, we found that the proposed technique achieved promising results under a variety of settings. Our experimental results also indicated that the proposed ULR technique significantly surpassed the other regular approaches in literature. Future work will address the issues of duplicate human names and explore supervised/semi-supervised learning techniques to further enhance the label quality with affordable human manual refinement efforts.