Robust Face-Name Graph Matching Essay Example for Free

Robust Face-Name Graph Matching Essay 1. Login In this module is going to explain the Robust Face-Name Graph Matching for Movie Character Identification designing and how we did the face detection and recognition in this project. The images will explain about the facial fetching details. After that admin going to login with the details which needed for the login page. 2. Detection In this module we are going to detect the face of the movie characters. In this module we are using the emgu cv library we must install the emgu cv library. After installing the emgu cv lib in our project we need to add reference with the name emgu.cv, emgu.cv.util, emgu.cv.ui. When you will complete the references you will get the emgu controls in the toolbox. 3. Recognition In this module we are going to recognize the face of the movie characters which is we previously stored on the face database. We just found that the give the real name of it. This is going to be done here. Here we are using the With the help of these eigenObjectRecognizer we are going to recognize the face. Chellangess In This Methodology :- 1. We detect the face in minute this is a big challenge for us because exiting system take more time for detection. 2. More challenging problem due to the huge variation in the appearance of each character. 3. It is increase speed of matching character and identify the character. Objectivies :- The Robust Face-Name Graph Matching for Movie Character Identification designing and how we did the face detection and recognition in this project. In this project we present two schemes of global face-name matching based framework for robust character identification. It is use in movies, video, cartoons. Problem Analysis :- 1. It is difficult to Complex character changes are handled by simultaneously graph partition and graph matching. 2. Many character are not easily matching and identification face in movies. 3. It is hard to be increase speed of matching character and identify the character. Existing Work :- This project is used to detect the face of movie characters and recognize the characters in minute process and the existing system are taking the too much time to detect the face. But this one we can do it in a minute process. Proposed Work :- In this Robust Face-Name Graph Matching for Movie Character Identification is used to detect the face of movie characters and the Proposed system is taking the minimum time to detect the face. In this One we can do it in a minute process. Robust Face-Name Graph Matching for Movie Character Identification Jitao Sang, Changsheng Xu, Senior Member, IEEE Abstract—Automatic face identification of characters in movies has drawn significant research interests and led to many interesting applications. It is a challenging problem due to the huge variation in the appearance of each character. Although existing methods demonstrate promising results in clean environment, the performances are limited in complex movie scenes due to the noises generated during the face tracking and face clustering process. In this paper we present two schemes of global face-name matching based framework for robust character identification. The contributions of this work include: 1) A noise insensitive character relationship representation is incorporated. 2) We introduce an edit operation based graph matching algorithm. 3) Complex character changes are handled by simultaneously graph partition and graph matching. 4) Beyond existing character identification approaches, we further perform an in-depth sensitivity analysis by introducing two types of simulated noises. The proposed schemes demonstrate state-of-the-art performance on movie character identification in various genres of movies. Index Terms—Character identification, graph matching, graph partition, graph edit, sensitivity analysis. Fig. 1. Examples of character identification from movie â€Å"Notting Hill†. I. INTRODUCTION A. Objective and Motivation The proliferation of movie and TV provides large amount of digital video data. This has led to the requirement of efficient and effective techniques for video content understanding and organization. Automatic video annotation is one of such key techniques. In this paper our focus is on annotating characters in the movie and TVs, which is called movie character identification [1]. The objective is to identify the faces of the characters in the video and label them with the corresponding names in the cast. The textual cues, like cast lists, scripts, subtitles and closed captions are usually exploited. Fig.1 shows an example in our experiments. In a movie, characters are the focus center of interests for the audience. Their occurrences provide lots of clues about the movie structure and content. Automatic character identification is essential for semantic movie index and retrieval [2], [3], scene segmentation [4], summarization [5] and other applications [6]. Copyright (c) 2010 IEEE. Personal use of this material is permitted. However, permission to use this material for any other purposes must be obtained from the IEEE by sending a request to pubs [emailprotected] ieee.org. This work was supported in part by the National Program on Key Basic Research Project (973 Program, Project No. 2012CB316304) and National Natural Science Foundation of China (Grant No. 90920303, 61003161). J. Sang and C. Xu (corresponding author) are with the National Lab of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences, Beijing 100190, China; and also with the China- Singapore Institute of Digital Media, Singapore, 119613 Character identification, though very intuitive to humans, is a tremendously challenging task in computer vision. The reason is four-fold: 1) Weakly supervised textual cues [7]. There are ambiguity problem in establishing the correspondence between names and faces: ambiguity can arise from a reaction shot where the person speaking may not be shown in the frames 1; ambiguity can also arise in partially labeled frames when there are multiple speakers in the same scene 2. 2) Face identification in videos is more difficult than that in images [8]. Low resolution, occlusion, no rigid deformations, large motion, complex background and other uncontrolled conditions make the results of face detection and tracking unreliable. In movies, the situation is even worse. This brings inevitable noises to the character identification. 3) The same character appears quite differently during the movie [3]. There may be huge pose, expression and illumination variation, wearing, clothing, even makeup and hairstyle changes. Moreover, characters in some movies go through different age stages, e.g., from youth to the old age. Sometimes, there will even be different actors playing different ages of the same character. 4) The determination for the number of identical faces is not trivial [2]. Due to the remarkable intra-class variance, the same character name will correspond to faces of huge variant appearances. It will be unreasonable to set the number of identical faces just according to the number of characters in the cast. Our study is motivated by these challenges and aims to find solutions for a robust framework for movie character identification. B. Related Work The crux of the character identification problem is to exploit the relations between videos and the associated texts in order 1 I.e., the name in the subtitle/closed caption finds no corresponding faces in the video. 2 I.e., multiple names in the subtitle/closed caption correspond to multiple faces in the video. Fig. 2. Framework of scheme 1: Face-name graph matching with #cluster pre specified to label the faces of characters with names. It has similarities to identifying faces in news videos [9], [10], [11]. However, in news videos, candidate names for the faces are available from the simultaneously appearing captions or local transcripts. While in TV and movies, the names of characters are seldom directly shown in the subtitle or closed caption, and script screenplay containing character names has no time stamps to align to the video. According to the utilized textual cues, we roughly divide the existing movie character identification methods into three categories. 1) Category 1: Cast list based: These methods only utilize the case list textual resource. In the â€Å"cast list discovery† problem [12], [13], faces are clustered by appearance and faces of a particular character are expected to be collected in a few pure clusters. Names for the clusters are then manually selected from the cast list. Ramanan et al. proposed to manually label an initial set of face clu sters and further cluster the rest face instances based on clothing within scenes [14]. In [15], the authors have addressed the problem of finding particular characters by building a model/classifier of the character’s appearance from user-provided training data. An interesting work combining character identification with web image retrieval is proposed in [17]. The character names in the cast are used as queries to search face images and constitute gallery set. The probe face tracks in the movie are then identified as one of the characters by multi-task joint sparse representation and classification. Recently, metric learning is introduced into character identification in uncontrolled videos [16]. Cast-specific metrics are adapted to the people appearing in a particular video in an unsupervised manner. The clustering as well as identification performance are demonstrated to be improved. These cast list based methods are easy for understanding and implementation. However, without other textual cues, they either need manual labeling or guarantee no robust clustering and classification performance due to the large intra-class variances. 2) Category 2: Subtitle or Closed caption, Local matching based: Subtitle and closed caption provide time-stamped dialogues, which can be exploited for alignment to the video frames. Effingham et al. [18], [3] proposed to combine the film script with the subtitle for local face-name matching. Time-stamped name annotation and face exemplars are generated. The rest of the faces were then classified into these exemplars for identification. They further extended their work in [19], by replacing the nearest neighbor classifier by multiple kernel learning for features combination. In the new framework, non-frontal faces are handled and the coverage is extended. Researchers from University of Pennsylvania utilized the readily available time-stamped resource, the closed captions, which is demonstrated more reliable than OCR-based subtitles [20], [7]. They investigated on the ambiguity issues in the local alignment between video, screenplay and closed captions. A partially-supervised multiclass classification problem is formulated. Recently, they attempted to address the character identification problem without the use of screenplay [21]. The reference cues in the closed captions are employed as multiple instance constraints and face tracks grouping as well as face-name association are solved in a convex formulation. The local matching based methods require the time-stamped information, which is either extracted by OCR (i.e., subtitle) or unavailable for the majority of movies and TV series (i.e., closed caption). Besides, the ambiguous and partial annotation makes local matching based methods more sensitive to the face detection and tracking noises. 3) Category 3: Script/Screenplay, Global matching based: Global matching based methods open the possibility of character identification without OCR-based subtitle or closed caption. Since it is not easy to get local name cues, the task of character identification is formulated as a global matching problem in [2], [22], [4]. Our method belongs to this category and can be considered as an extension to Zhang’s work [2]. In movies, the names of characters seldom directly appear in the subtitle, while the movie script which contains character names has no time information. Without the local time information, the task of character identification is formulated as a global matching problem between the faces detected from the video and the names extracted from the movie script. Compared with local matching, global statistics are used for name-face association, which enhances the robustness of the algorithms. Our work differs from the existing research in threefold: Regarding the fact that characters may show various appearances, the representation of character is often affected Fig. 3. Framework of scheme 2: Face-name graph matching without #cluster pre-specified. by the noise introduced by face tracking, face clustering and scene segmentation. Although extensive research efforts have been concentrated on character identification and many applications have been proposed, little work has focused on improving the robustness. We have observed in our investigations that some statistic properties are preserved in spite of these noises. Based on that, we propose a novel representation for character relationship and introduce a name-face matching method which can accommodate a certain noise. Face track clustering serves as an important step in movie character identification. In most of the existing methods some cues are utilized to determine the number of target clusters prior to face clustering, e.g., in [2], the number of clusters is the same as the number of distinct speakers appearing in the script. While this seems convinced at first glance, it is rigid and even deteriorating the clustering results sometimes. In this paper, we loose the restriction of one face cluster corresponding to one character name. Face track clustering and face-name matching are jointly optimized and conducted in a unique framework. Sensitivity analysis is common in financial applications, risk analysis, signal processing and any area where models are developed [23], [24]. Good modeling practice requires that the modeler provides an evaluation of the confidence in the model, for example, assessing the uncertainties associated with the modeling process and with the outcome of the model itself. For movie character identification, sensitivity analysis offers valid tools for characterizing the robustness to noises for a model. To the best of our knowledge, there have been no efforts directed at the sensitivity analysis for movie character identification. In this paper, we aim to fill this gap by introducing two types of simulated noises. A preliminary version of this work was introduced by [1]. We provide additional algorithmic and computational details, and extend the framework considering no pre-specification for the number of face clusters. Improved performance as well as robustness are demonstrated in movies wit h large character appearance changes. C. Overview of Our Approach In this paper, we propose a global face-name graph matching based framework for robust movie character identification. Two schemes are considered. There are connections as well as differences between them. Regarding the connections, firstly, the proposed two schemes both belong to the global matching based category, where external script resources are utilized. Secondly, to improve the robustness, the ordinal graph is employed for face and name graph representation and a novel graph matching algorithm called Error Correcting Graph Matching (ECGM) is introduced. Regarding the differences, scheme 1 sets the number of clusters when performing face clustering (e.g., K-means, spectral clustering). The face graph is restricted to have identical number of vertexes with the name graph. While, in scheme 2, no cluster number is required and face tracks are clustered based on their intrinsic data structure (e.g., mean shift, affinity propagation). Moreover, as shown in Fig.2 and Fig.3, scheme 2 has an additional module of graph partition compared with scheme 1. From this perspective, scheme 2 can be seen as an extension to scheme 1. m1) Scheme 1: The proposed framework for scheme 1 is shown in Fig.2. It is similar to the framework of [2]. Face tracks are clustered using constrained K-means, where the number of clusters is set as the number of distinct speakers. Co-occurrence of names in script and face clusters in video constitutes the corresponding face graph and name graph. We modify the traditional global matching framework by using ordinal graphs for robust representation and introducing an ECGM-based graph matching method. For face and name graph construction, we propose to represent the character co-occurrence in rank ordinal level [25], which scores thestrength of the relationships in a rank order from the weakest to strongest. Rank order data carry no numerical meaning and thus are less sensitive to the noises. The affinity graph used in the traditional global matching is interval measures of the co-occurrence relationship between characters. While continuous measures of the strength of relationship holds complete information, it is highly sensitive to noises. For name-face graph matching, we utilize the ECGM algorithm. In ECGM, the difference between two graphs is measured by edit distance which is a sequence of graph edit operations. The optimal match is achieved with the least edit distance. According to the noise analysis, we define appropriate graph edit operations and adapt the distance functions to obtain improved name-face matching performance. 2) Scheme 2: The proposed framework for scheme 2 is shown in Fig.3. It has two differences from scheme 1 in Fig.2. First, no cluster number is required for the face tracks clustering step. Second, since the face graph and name graph may have different number of vertexes, a graph partition component is added before ordinal graph representation. The basic premise behind the scheme 2 is that appearances of the same character vary significantly and it is difficult to group them in a unique cluster. Take the movie â€Å"TheCurious Case of Benjamin Button† for example. The hero and heroine go through a long time period from their childhood, youth, middle-age to the old-age. The intra-class variance is even larger than the inter-class variance. In this case, simply enforcing the number of face clusters as the numberof characters will disturb the clustering process. Instead of grouping face tracks of the same character into one cluster, face tracks from different characters may be grouped together. In scheme 2, we utilize affinity propagation for the face tracks clustering. With each sample as the potential center of clusters, the face tracks are recursively clustered through appearance-based similarity transmit and propagation. High cluster purity with large number of clusters is expected. Since one character name may correspond to several face clusters, graph partition is introduced before graph matching. Which face clusters should be further grouped (i.e., divided into th e same subgraph) is determined by whether the partitioned face graph achieves an optimal graph matching with the name graph. Actually, face clustering is divided into two steps: coarse clustering by appearance and further modification by script. Moreover, face clustering and graph matching are optimized simultaneously, which improve the robustness against errors and noises. In general, the scheme 2 has two advantages over the scheme 1. (a) For scheme 2, no cluster number is required in advance and face tracks are clustered based on their intrinsic data structure. Therefore, the scheme 2 provides certain robustness to the intra-class variance, which is very common in movies where characters change appearance significantly or go through a long time period. (b) Regarding that movie cast cannot include pedestrians whose face is detected and added into the face track, restricting the number of face tracks clusters the same as that of name from movie cast will deteriorate the clustering process. In addition, there is some chance that movie cast does not cover all the characters. In this case, pre-specification for the face clusters is risky: face tracks from different characters will be mixed together and graph matching tends to fail. 3) Sensitivity Analysis: Sensitivity analysis plays an important role in characterizing the uncertainties associated with a model. To explicitly analyze the algorithm’s sensitivity to noises, two types of noises, coverage noise and intensity noise, are introduced. Based on that, we perform sensitivity analysis by investigating the performance of name-face matching with respect to the simulated noises.