The explosion of image data on the Internet has the potential to foster more sophisticated and robust models and algorithms to index, retrieve, organize and interact with images and multimedia data. But exactly how such data can be harnessed and organized remains a critical problem. We introduce here a new database called “ImageNet”, a large-scale ontology of images built upon the backbone of the WordNet structure. ImageNet aims to populate the majority of the 80,000 synsets of WordNet with an average of 500-1000 clean and full resolution images. This will result in tens of millions of annotated images organized by the semantic hierarchy of WordNet. This paper offers a detailed analysis of ImageNet in its current state: 12 subtrees with 5247 synsets and 3.2 million images in total. We show that ImageNet is much larger in scale and diversity and much more accurate than the current image datasets. Constructing such a large-scale database is a challenging task. We describe the data collection scheme with Amazon Mechanical Turk. Lastly, we illustrate the usefulness of ImageNet through three simple applications in object recognition, image classification and automatic object clustering. We hope that the scale, accuracy, diversity and hierarchical structure of ImageNet can offer unparalleled opportunities to researchers in the computer vision community and beyond.

/pdf/imagenet-a-large-scale-hierarchical-image-database-1e8f663s1h.pdf

ImageNet: A large-scale hierarchical image database

This paper presents a method for extracting distinctive invariant features from images that can be used to perform reliable matching between different views of an object or scene. The features are invariant to image scale and rotation, and are shown to provide robust matching across a substantial range of affine distortion, change in 3D viewpoint, addition of noise, and change in illumination. The features are highly distinctive, in the sense that a single feature can be correctly matched with high probability against a large database of features from many images. This paper also describes an approach to using these features for object recognition. The recognition proceeds by matching individual features to a database of features from known objects using a fast nearest-neighbor algorithm, followed by a Hough transform to identify clusters belonging to a single object, and finally performing verification through least-squares solution for consistent pose parameters. This approach to recognition can robustly identify objects among clutter and occlusion while achieving near real-time performance.

/pdf/distinctive-image-features-from-scale-invariant-keypoints-3waurjqzke.pdf

Distinctive Image Features from Scale-Invariant Keypoints

Fiji is a distribution of the popular open-source software ImageJ focused on biological-image analysis. Fiji uses modern software engineering practices to combine powerful software libraries with a broad range of scripting languages to enable rapid prototyping of image-processing algorithms. Fiji facilitates the transformation of new algorithms into ImageJ plugins that can be shared with end users through an integrated update system. We propose Fiji as a platform for productive collaboration between computer science and biology research communities.

/pdf/fiji-an-open-source-platform-for-biological-image-analysis-1v529iu3cx.pdf

Fiji: an open-source platform for biological-image analysis

We study the question of feature sets for robust visual object recognition; adopting linear SVM based human detection as a test case. After reviewing existing edge and gradient based descriptors, we show experimentally that grids of histograms of oriented gradient (HOG) descriptors significantly outperform existing feature sets for human detection. We study the influence of each stage of the computation on performance, concluding that fine-scale gradients, fine orientation binning, relatively coarse spatial binning, and high-quality local contrast normalization in overlapping descriptor blocks are all important for good results. The new approach gives near-perfect separation on the original MIT pedestrian database, so we introduce a more challenging dataset containing over 1800 annotated human images with a large range of pose variations and backgrounds.

/pdf/histograms-of-oriented-gradients-for-human-detection-3nteinziuq.pdf

Histograms of oriented gradients for human detection

The ImageNet Large Scale Visual Recognition Challenge is a benchmark in object category classification and detection on hundreds of object categories and millions of images. The challenge has been run annually from 2010 to present, attracting participation from more than fifty institutions. This paper describes the creation of this benchmark dataset and the advances in object recognition that have been possible as a result. We discuss the challenges of collecting large-scale ground truth annotation, highlight key breakthroughs in categorical object recognition, provide a detailed analysis of the current state of the field of large-scale image classification and object detection, and compare the state-of-the-art computer vision accuracy with human accuracy. We conclude with lessons learned in the 5 years of the challenge, and propose future directions and improvements.

/pdf/imagenet-large-scale-visual-recognition-challenge-caky5nmmix.pdf

ImageNet Large Scale Visual Recognition Challenge

A method and apparatus for identifying scale invariant features in an image and a further method and apparatus for using such scale invariant features to locate an object in an image are disclosed. The method and apparatus for identifying scale invariant features may involve the use of a processor circuit for producing a plurality of component subregion descriptors for each subregion of a pixel region about pixel amplitude extrema in a plurality of difference images produced from the image. This may involve producing a plurality of difference images by blurring an initial image to produce a blurred image and by subtracting the blurred image from the initial image to produce the difference image. For each difference image, pixel amplitude extrema are located and a corresponding pixel region is defined about each pixel amplitude extremum. Each pixel region is divided into subregions and a plurality of component subregion descriptors are produced for each subregion. These component subregion descriptors are correlated with component subregion descriptors of an image under consideration and an object is indicated as being detected when a sufficient number of component subregion descriptors (scale invariant features) define an aggregate correlation exceeding a threshold correlation with component subregion descriptors (scale invariant features) associated with the object.

Method and apparatus for identifying scale invariant features in an image and use of same for locating an object in an image

We investigate the role of sparsity and localized features in a biologically-inspired model of visual object classification. As in the model of Serre, Wolf, and Poggio, we first apply Gabor filters at all positions and scales; feature complexity and position/scale invariance are then built up by alternating template matching and max pooling operations. We refine the approach in several biologically plausible ways. Sparsity is increased by constraining the number of feature inputs, lateral inhibition, and feature selection. We also demonstrate the value of retaining some position and scale information above the intermediate feature level. Our final model is competitive with current computer vision algorithms on several standard datasets, including the Caltech 101 object categories and the UIUC car localization task. The results further the case for biologically-motivated approaches to object classification.

/pdf/object-class-recognition-and-localization-using-sparse-1gxc239c59.pdf

Object Class Recognition and Localization Using Sparse Features with Limited Receptive Fields

A computer vision system has been developed for real-time motion tracking of 3-D objects, including those with variable internal parameters. This system provides for the integrated treatment of matching and measurement errors that arise during motion tracking. These two sources of error have very different distributions and are best handled by separate computational mechanisms. These errors can be treated in an integrated way by using the computation of variance in predicted feature measurements to determine the probability of correctness for each potential matching feature. In return, a best-first search procedure uses these probabilities to find consistent sets of matches, which eliminates the need to treat outliers during the analysis of measurement errors. The most reliable initial matches are used to reduce the parameter variance on further iterations, minimizing the amount of search required for matching more ambiguous features. These methods allow for much larger frame-to-frame motions than most previous approaches. The resulting system can robustly track models with many degrees of freedom while running on relatively inexpensive hardware. These same techniques can be used to speed verification during model-based recognition.

/pdf/robust-model-based-motion-tracking-through-the-integration-3g05di5vw4.pdf

Robust Model-based Motion Tracking Through the Integration of Search and Estimation

We present a complete system architecture for fully automated markerless augmented reality (AR). The system constructs a sparse metric model of the real-world environment, provides interactive means for specifying the pose of a virtual object, and performs model-based camera tracking with visually pleasing augmentation results. Our approach does not require camera pre-calibration, prior knowledge of scene geometry, manual initialization of the tracker or placement of special markers. Robust tracking in the presence of occlusions and scene changes is achieved by using highly distinctive natural features to establish image correspondences.

/pdf/scene-modelling-recognition-and-tracking-with-invariant-1n047b0w6c.pdf

Scene Modelling, Recognition and Tracking with Invariant Image Features

There has been growing interest in the use of binary-valued features, such as BRIEF, ORB, and BRISK for efficient local feature matching. These binary features have several advantages over vector-based features as they can be faster to compute, more compact to store, and more efficient to compare. Although it is fast to compute the Hamming distance between pairs of binary features, particularly on modern architectures, it can still be too slow to use linear search in the case of large datasets. For vector-based features, such as SIFT and SURF, the solution has been to use approximate nearest-neighbor search, but these existing algorithms are not suitable for binary features. In this paper we introduce a new algorithm for approximate matching of binary features, based on priority search of multiple hierarchical clustering trees. We compare this to existing alternatives, and show that it performs well for large datasets, both in terms of speed and memory efficiency.

/pdf/fast-matching-of-binary-features-27zxz90seo.pdf

David G. Lowe

Papers

Method and apparatus for identifying scale invariant features in an image and use of same for locating an object in an image

Object Class Recognition and Localization Using Sparse Features with Limited Receptive Fields

Robust Model-based Motion Tracking Through the Integration of Search and Estimation

Scene Modelling, Recognition and Tracking with Invariant Image Features

Fast Matching of Binary Features