Identifying relevant frames in weakly labeled videos for training concept detectors
Published 7 July 2008
Adrian Ulges, Christian Schulze, Daniel Keysers, Thomas M. Breuel
Citations44
Generate an AI Snapshot to get a quick, structured summary of this paper.
Study Snapshot
ObjectiveStudy objective
MethodsResearch methodology
PopulationPopulation studied
Sample sizeSample sizes
OutcomesStudy outcomes here
ResultsStudy results comes here
LimitationsResearch study limitations comes here
A concise AI-generated summary of the paper will appear here once you click Generate AI Snapshot.
TL;DR
A probabilistic framework for learning from weakly annotated training videos in the presence of irrelevant content is presented, and the relevance of keyframes is modeled as a latent random variable that is estimated during training.
Abstract
A key problem with the automatic detection of semantic concepts (like 'interview' or 'soccer') in video streams is the manual acquisition of adequate training sets. Recently, we have proposed to use online videos downloaded from portals like youtube.com for this purpose, whereas tags provided by users during video upload serve as ground truth annotations.
Keywords
Computer Science
Journal of the Royal Statistical Society Series B (Statistical Methodology)Maximum Likelihood from Incomplete Data Via the <i>EM</i> Algorithm
49,657 Citations1977A. P. Dempster, N. M. Laird +1 more
Beyond Bags of Features: Spatial Pyramid Matching for Recognizing Natural Scene Categories
7,929 Citations2006Svetlana Lazebnik, C. Schmid +1 more
This paper presents a method for recognizing scene categories based on approximate global geometric correspondence that exceeds the state of the art on the Caltech-101 database and achieves high accuracy on a large database of fifteen natural scene categories.
Advances in industrial controlPattern Classification
7,884 Citations2000Evan L. Russell, Leo H. Chiang +1 more
IEEE Transactions on Pattern Analysis and Machine IntelligenceA performance evaluation of local descriptors
6,692 Citations2005Krystian Mikolajczyk, C. Schmid
Machine LearningUnsupervised Learning by Probabilistic Latent Semantic Analysis
2,443 Citations2001Thomas Hofmann
This paper proposes to make use of a temperature controlled version of the Expectation Maximization algorithm for model fitting, which has shown excellent performance in practice, and results in a more principled approach with a solid foundation in statistical inference.
IEEE Transactions on Systems Man and CyberneticsTextural Features Corresponding to Visual Perception
2,130 Citations1978Hideyuki Tamura, Shunji Mori +1 more
The discrepancies between human vision and computerized techniques that are encountered in this study indicate fundamental problems in digital analysis of textures and could be overcome by analyzing their causes and using more sophisticated techniques.
Object class recognition by unsupervised scale-invariant learning
2,035 Citations2003Rob Fergus, Pietro Perona +1 more
The flexible nature of the model is demonstrated by excellent results over a range of datasets including geometrically constrained classes (e.g. faces, cars) and flexible objects (such as animals).
International Journal of Computer VisionLocal Features and Kernels for Classification of Texture and Object Categories: A Comprehensive Study
1,931 Citations2006Jianguo Zhang, M. Marszałek +2 more
A large-scale evaluation of an approach that represents images as distributions of features extracted from a sparse set of keypoint locations and learns a Support Vector Machine classifier with kernels based on two effective measures for comparing distributions, the Earth Mover’s Distance and the χ2 distance.
International Journal of Computer VisionEvaluation of Interest Point Detectors
1,441 Citations2000Cordelia Schmid, Roger Mohr +1 more
Two evaluation criteria for interest points' repeatability rate and information content are introduced and different interest point detectors are compared using these two criteria.
Discovering objects and their location in images
980 Citations2005Josef Šivic, Bryan Russell +3 more
This work treats object categories as topics, so that an image containing instances of several categories is modeled as a mixture of topics, and develops a model developed in the statistical text literature: probabilistic latent semantic analysis (pLSA).
A performance evaluation of local descriptors
952 Citations2003Krystian Mikolajczyk, C. Schmid
It is observed that the ranking of the descriptors does not depend on the point detector and that SIFT descriptors perform best and steerable filters can be considered a good choice given the low dimensionality.
Learning object categories from Google's image search
708 Citations2005Rob Fergus, Li Fei-Fei +2 more
A new model, TSI-pLSA, is developed, which extends pLSA (as applied to visual words) to include spatial information in a translation and scale invariant manner, and can handle the high intra-class variability and large proportion of unrelated images returned by search engines.
IEEE Transactions on Pattern Analysis and Machine IntelligenceGeneric object recognition with boosting
372 Citations2006Andreas Opelt, Axel Pinz +2 more
This paper presents a complete framework that starts with the extraction of various local regions of either discontinuity or homogeneity, and uses Boosting to learn a subset of feature vectors (weak hypotheses) and to combine them into one final hypothesis for each visual category.
Image database retrieval with multiple-instance learning techniques
160 Citations2002Yang Chen, Tomás Lozano‐Pérez
An approach for retrieving images from an image database based on content similarity using a multiple-instance learning method known as the diverse density algorithm, which tries to retrieve images with similar feature vectors from the remainder of the database.
IBM Research TRECVID-2006 Video Retrieval System
100 Citations2006Murray Campbell, Alexander Haubold +7 more
Discriminative Training for Object Recognition Using Image Patches
91 Citations2005Thomas Deselaers, Daniel Keysers +1 more
A method for automatically learning discriminative image patches for the recognition of given object classes is presented and decides that patches containing an eye are most important for distinguishing face from background images.
Lecture notes in computer scienceA System That Learns to Tag Videos by Watching Youtube
34 Citations2008Adrian Ulges, Christian Schulze +2 more
A system that automatically tags videos, i.e. detects high-level semantic concepts like objects or actions in them, that uses videos from online portals like youtube.com as a novel source of training data, whereas tags provided by users during upload serve as ground truth annotations, to learn autonomously by automatically downloading its training set.
Training Object Detection Models with Weakly Labeled Data
8 Citations2002Charles Rosenberg, M. Hebert
This work describes a method whereby a training set consisting of a small number of fully labeled training examples augmented with a set of weakly labeled examples can be used to train a detector which exhibits performance better than that which can be obtained with a reduced set of fully labeling training examples alone.