Search

Deep learning algorithms are a subset of the machine learning algorithms, which aim at discovering multiple levels of distributed representations. Recently, numerous deep learning algorithms have been proposed to solve traditional artificial intelligence problems. This work aims to review the state-of-the-art in deep learning algorithms in computer vision by highlighting the contributions and challenges from over 210 recent research papers. It first gives an overview of various deep learning approaches and their recent developments, and then briefly describes their applications in diverse vision tasks, such as image classification, object detection, image retrieval, semantic segmentation and human pose estimation. Finally, the paper summarizes the future trends and challenges in designing and training deep neural networks.

Image-to-image translation between different domains has been an important research direction, with the aim of arbitrarily manipulating the source image content to become similar to a target image. Recently, cycle-consistent generative network (CycleGAN) has become a fundamental approach for general-purpose image-to-image translation, while almost no work has examined what factors may influence its performance. To provide more insights, we propose two new models roughly based on CycleGAN, namely Long CycleGAN and Nest CycleGAN. First, Long CycleGAN cascades several generators to perform the domain translation in a long cycle. It shows the benefit of stacking more generators on the generation quality. In addition to the long cycle, Nest CycleGAN develops new inner cycles to bridge intermediate generators directly, which can help constrain the unsupervised mappings. In the experiments, we conduct qualitative and quantitative comparisons for tasks including photo label, photo sketch, and photo colorization. The quantitative and qualitative results demonstrate the effectiveness of our two proposed models.

Computational Fluid Dynamics (CFD) simulations are able to produce complex and large outputs that accurately describe the physical properties of fluids and gases in various domains, such as air flow around a car, or the multi-phase flow inside an internal combustion engine. The simulation results, i.e. the flow fields, are often too complex to be analyzed directly. With the increasing number of simulations as well as their complexity, there is a need of automated processes that can analyze these complex outputs. In this paper, inspired by the success of convolutional neural networks (CNNs) in Computer Vision, we apply for the first time CNNs on CFD output. We show their capabilities in capturing and processing flow patterns. Furthermore, we design a novel CNN architecture tailored to the data produced by CFD simulations, as well as two conventional architectures and compare them. We propose and construct a new dataset of turbulent flow, within the application domain of steady flow around passenger cars. We use that dataset to evaluate and compare the proposed methods, on different tasks that depend on flow patterns. Finally, we compare our methods with a baseline k-nearest neighbor approach, tuned to be comparable to the state-ofthe-art.

In recent years, deep learning has been successfully applied to diverse multimedia research areas, with the aim of learning powerful and informative representations for a variety of visual recognition tasks. In this work, we propose convolutional fusion networks (CFN) to integrate multi-level deep features and fuse a richer visual representation. Despite recent advances in deep fusion networks, they still have limitations due to expensive parameters and weak fusion modules. Instead, CFN uses 1 × 1 convolutional layers and global average pooling to generate side branches with few parameters, and employs a locally-connected fusion module, which can learn adaptive weights for different side branches and form a better fused feature. Specifically, we introduce three key components of the proposed CFN, and discuss its differences from other deep models. Moreover, we propose fully convolutional fusion networks (FCFN) that are an extension of CFN for pixel-level classification applied to several tasks, such as semantic segmentation and edge detection. Our experiments demonstrate that CFN (and FCFN) can achieve promising performance by consistent improvements for both image-level and pixel-level classification tasks, compared to a plain CNN. We release our codes on https://github.com/yuLiu24/CFN . Also, we make a live demo ( goliath.liacs.nl ) using a CFN model trained on the ImageNet dataset.

Objects are often organized in a semantic hierarchy of categories, where fine-level categories are grouped into coarse-level categories according to their semantic relations. While previous works usually only classify objects into the leaf categories, we argue that generating hierarchical labels can actually describe how the leaf categories evolved from higher level coarse-grained categories, thus can provide a better understanding of the objects. In this paper, we propose to utilize the CNN-RNN framework to address the hierarchical image classification task. CNN allows us to obtain discriminative features for the input images, and RNN enables us to jointly optimize the classification of coarse and fine labels. This framework can not only generate hierarchical labels for images, but also improve the traditional leaf-level classification performance due to incorporating the hierarchical information. Moreover, this framework can be built on top of any CNN architecture which is primarily designed for leaf-level classification. Accordingly, we build a high performance network based on the CNN-RNN paradigm which outperforms the original CNN (wider-ResNet) and also the current state-of-the-art. In addition, we investigate how to utilize the CNN-RNN framework to improve the fine category classification when a fraction of the training data is only annotated with coarse labels. Experimental results demonstrate that CNN-RNN can use the coarse-labeled training data to improve the classification of fine categories, and in some cases it even surpasses the performance achieved by fully annotated training data. This reveals that, CNN-RNN can alleviate the challenge of specialized and expensive annotation of fine labels.

Multimodal learning has been an important and challenging problem for decades, which aims to bridge the modality gap between heterogeneous representations, such as vision and language. Unlike many current approaches which only focus on either multimodal matching or classification, we propose a unified network to jointly learn multimodal matching and classification (MMC-Net) between images and texts. The proposed MMC-Net model can seamlessly integrate the matching and classification components. It first learns visual and textual embedding features in the matching component, and then generates discriminative multimodal representations in the classification component. Combining the two components in a unified model can help in improving their performance. Moreover, we present a multi-stage training algorithm by minimizing both of the matching and classification loss functions. Experimental results on four well-known multimodal benchmarks demonstrate the effectiveness and efficiency of the proposed approach, which achieves competitive performance for multimodal matching and classification compared to state-of-the-art approaches.

Convolutional neural networks (CNNs) have attracted significant attention in visual recognition. Several recent studies have shown that, in addition to the fully connected layers, the features derived from the convolutional layers of CNNs can also achieve promising performance in image classification tasks. In this paper, we propose a new feature from the convolutional layers, called Bag of Surrogate Parts (BoSP), and its spatial variant, Spatial-BoSP (S-BoSP). The main idea is, we assume the feature maps in the convolutional layers as surrogate parts, and densely sample and assign image regions to these surrogate parts by observing the activation values. Together with BoSP/S-BoSP, we further propose another two schemes to enhance the performance: scale pooling and global-part prediction. Scale pooling aims to handle the objects with different scales and deformations, and global-part prediction combines the predictions of global and part features. By conducting extensive experiments on generic object, fine-grained object and scene datasets, we find the proposed scheme can not only achieve superior performance to the fully connected feature, but also produces competitive or, in some cases, remarkably better performance than the state of the art.

General captioning practice involves a single forward prediction, with the aim of predicting the word in the next timestep given the word in the current timestep. In this paper, we present a novel captioning framework, namely Dual Prediction Network (DPN), which is end-to-end trainable and addresses the captioning problem with dual predictions. Specifically, the dual predictions consist of a forward prediction to generate the next word from the current input word, as well as a backward prediction to reconstruct the input word using the predicted word. DPN has two appealing properties: 1) By introducing an extra supervision signal on the prediction, DPN can better capture the interplay between the input and the target; 2) Utilizing the reconstructed input, DPN can make another new prediction. During the test phase, we average both predictions to formulate the final target sentence. Experimental results on the MS COCO dataset demonstrate that, benefiting from the reconstruction step, both generated predictions in DPN outperform the predictions of methods based on the general captioning practice (single forward prediction), and averaging them can bring a further accuracy boost. Overall, DPN achieves competitive results with state-of-the-art approaches, across multiple evaluation metrics. © 2018 IEEE.

A major challenge in matching between vision and language is that they typically have completely different features and representations. In this work, we introduce a novel bridge between the modality-specific representations by creating a co-embedding space based on a recurrent residual fusion (RRF) block. Specifically, RRF adapts the recurrent mechanism to residual learning, so that it can recursively improve feature embeddings while retaining the shared parameters. Then, a fusion module is used to integrate the intermediate recurrent outputs and generates a more powerful representation. In the matching network, RRF acts as a feature enhancement component to gather visual and textual representations into a more discriminative embedding space where it allows to narrow the crossmodal gap between vision and language. Moreover, we employ a bi-rank loss function to enforce separability of the two modalities in the embedding space. In the experiments, we evaluate the proposed RRF-Net using two multi-modal datasets where it achieves state-of-the-art results.

One challenging problem in semantic segmentation is due to the erroneous predictions between categorical foreground and cluttered background. To address it, we propose to utilize a fused loss function to train a fully convolutional network, which aims to enhance the discrimination between foreground and background in images. In addition, we propose a pixel objectness (POS) to measure the importance of pixels. POS is able to recover some missing foreground pixels from the background. Experimental results on the PASCAL VOC 2012 dataset demonstrate our approach can achieve considerable improvements compared with the baseline counterpart, while maintaining the ease of training deep networks.

As local detectors and descriptors can find and represent distinctive keypoints in an image, various types of keypoints detection and description methods have been proposed. Each method has particular advantages and limitations and may be appropriate in different contexts. In this paper, we evaluate the performance of a wide set of local detectors and descriptors. First, we compare diverse local detectors with regard to the repeatability, and local descriptors in terms of the recall and precision. Next, we apply the visual words model constructed from the local descriptors with real values and binary string to large scale image search. The evaluation results reveal some strengths and weaknesses of the recent binary string descriptors compared with the notable real valued descriptors. Finally, we integrate the local detectors and descriptors with the framework of fully affine space and evaluate their performance under major viewpoint transformations. The presented comparative experimental studies can support researchers in choosing an appropriate local detector and descriptor for their specific computer vision applications.

Nowadays, a general pipeline for the image captioning task takes advantage of image representations based on convolutional neural networks (CNNs) and sequence modeling based on recurrent neural networks (RNNs). As captioning performance closely depends on the discriminative capacity of CNNs, our work aims to investigate the effects of different Convnets (CNN models) on image captioning. We train three Convnets based on different classification tasks: single-label, multi-label and multi-attribute, and then feed visual representations from these Convnets into a Long Short-Term Memory (LSTM) to model the sequence of words. Since the three Convnets focus on different visual contents in one image, we propose aggregating them together to generate a richer visual representation. Furthermore, during testing, we use an efficient multi-scale augmentation approach based on fully convolutional networks (FCNs). Extensive experiments on the MS COCO dataset provide significant insights into the effects of Convnets. Finally, we achieve comparable results to the state-of-the-art for both caption generation and image-sentence retrieval tasks.

Despite recent advances in multi-scale deep representations, their limitations are attributed to expensive parameters and weak fusion modules. Hence, we propose an efficient approach to fuse multi-scale deep representations, called convolutional fusion networks (CFN). Owing to using 1 \(\times \) 1 convolution and global average pooling, CFN can efficiently generate the side branches while adding few parameters. In addition, we present a locally-connected fusion module, which can learn adaptive weights for the side branches and form a discriminatively fused feature. CFN models trained on the CIFAR and ImageNet datasets demonstrate remarkable improvements over the plain CNNs. Furthermore, we generalize CFN to three new tasks, including scene recognition, fine-grained recognition and image retrieval. Our experiments show that it can obtain consistent improvements towards the transferring tasks.

Wepropose using relaxed deep supervision (RDS) within convolutional neural networks for edge detection. The conventional deep supervision utilizes the general groundtruth to guide intermediate predictions. Instead, we build hierarchical supervisory signals with additional relaxed labels to consider the diversities in deep neural networks. We begin by capturing the relaxed labels from simple detectors (e.g. Canny). Then we merge them with the general groundtruth to generate the RDS. Finally we employ the RDS to supervise the edge network following a coarse-to-fine paradigm. These relaxed labels can be seen as some false positives that are difficult to be classified. Weconsider these false positives in the supervision, and are able to achieve high performance for better edge detection. Wecompensate for the lack of training images by capturing coarse edge annotations from a large dataset of image segmentations to pretrain the model. Extensive experiments demonstrate that our approach achieves state-of-the-art performance on the well-known BSDS500 dataset (ODS F-score of .792) and obtains superior cross-dataset generalization results on NYUD dataset.

Recent advances in image classification have been achieved with deep convolutional neural networks (CNNs). The pooling and sub-sampling operations in the CNNs introduce invariance to local transformations, but result in accuracy loss in the image applications. In this paper, we propose a novel deep network called “Weighted Integration Architecture Network” (WIAN) which can effectively recover the information loss due to the pooling operation in the CNNs. The proposed WIAN reuses the information from the previous layers in the network and assigns a weight to each according to the responses or entropy in the layer and then element-wise summing them to further enhance the image classification performance. Exhaustive experiments on four standard benchmark datasets (CIFAR-10, CIFAR-100, MNIST and SVHN) demonstrate the effectiveness as well as an improved performance of WIAN.

In this paper, we present a novel framework which significantly increases the accuracy of correspondences matching between two images under various image transformations. We first define a retina inspired patch-structure which mimics the human eye retina topology, and use the highly discriminative convolutional neural networks (CNNs) features to represent those patches. Then, we employ the conventional salient point methods to locate salient points, and finally, we fuse both the local descriptor of each salient point and the CNN feature from the local patch which the salient point belongs to. The evaluation results show the effectiveness of the proposed multiple features fusion (MFF) framework and that it improves the accuracy of leading approaches on two popular benchmark datasets.

We are living in an Age of Information where the amount of accessible data from science and culture is almost limitless. However, this also means that finding an item of interest is increasingly difficult, a digital needle in the prover- bial haystack. In this article, we focus on the topic of content- based image retrieval using interactive search techniques, i.e., how does one interactively find any kind of imagery from any source, regardless of whether it is photographic, MRI or X-ray? We highlight trends and ideas from over 170 recent research papers aiming to capture the wide spec- trum of paradigms and methods in interactive search, includ- ing its subarea relevance feedback. Furthermore, we identify promising research directions and several grand challenges for the future.

One of the grand challenges of machine intelligence and pattern recognition for the past decade has been bridging the semantic gap, that is, determining how to translate the low-level features from images, video and audio to the high-level concepts of humans. Concept detection is an important approach toward bridging the semantic gap by allowing computers to understand imagery using the conceptual vocabulary of humans. By exploiting big data, the current generation of algorithms has contributed and developed both advances in accuracy and computational efficiency as well as new paradigms and techniques in concept detection. This special issue provides a focus on the state-of-the-art in concept detection with big data. We received 21 submissions of which 16 were selected for the triple peer-review process. In this special issue, we are pleased to present six research papers on concept detection with big data that present the latest advances in this field

The features extracted from convolutional neural networks (CNNs) are able to capture the discriminative part of an image and have shown superior performance in visual recognition. Furthermore, it has been verified that the CNN activations trained from large and diverse datasets can act as generic features and be transferred to other visual recognition tasks. In this paper, we aim to learn more from an image and present an effective method called Principal Pyramidal Convolution (PPC). The scheme first partitions the image into two levels, and extracts CNN activations for each sub-region along with the whole image, and then aggregates them together. The concatenated feature is later reduced to the standard dimension using Principal Component Analysis (PCA) algorithm, generating the refined CNN feature. When applied in image classification and retrieval tasks, the PPC feature consistently outperforms the conventional CNN feature, regardless of the network type where they derive from. Specifically, PPC achieves state-of-the-art result on the MIT Indoor67 dataset, utilizing the activations from Places-CNN.

Extending beyond the boundaries of science, art, and culture, content-based multimedia information retrieval provides new paradigms and methods for searching through the myriad variety of media all over the world. This survey reviews 100+ recent articles on content-based multimedia information retrieval and discusses their role in current research directions which include browsing and search paradigms, user studies, affective computing, learning, semantic queries, new features and media types, high performance indexing, and evaluation techniques. Based on the current state of the art, we discuss the major challenges for the future.

We present the Retina-inspired Invariant Fast Feature, RIFF, which was designed for invariance to scale, rotation, and affine image deformations. The feature descriptor is based on pair-wise comparisons over a sampling pattern loosely based on the human retina and introduces a method for improving accuracy by maximizing the discriminatory power of the point set. A performance evaluation with regard to bag of words based image retrieval on several well-known international datasets demonstrates that the RIFF descriptor has competitive performance to the state-of-the-art descriptors (e.g. SIFT, SURF, BRISK, and FREAK).

Image copy detection is a major challenge with regard to computational efficiency, memory requirements and accuracy. A popular approach from the literature is to use visual words (or Bag-of-Words) constructed from real value (SIFT and SURF) and binary string salient point descriptors (BRIEF, ORB, BRISK and FREAK). To accommodate large scale data sets, we used the approximate nearest neighbor (ANN) based cluster approach. Our results on several well-known test sets reveal that some of the recent binary string approaches outperformed notable descriptors such as SIFT and SURF.

Deep learning has achieved significant attention recently due to promising results in representing and classifying concepts most prominently in the form of convolutional neural networks CNN. While CNN has been widely studied and evaluated in computer vision, there are other forms of deep learning algorithms which may be promising. One interesting deep learning approach which has received relatively little attention in visual concept classification is Cascade-Correlation Neural Networks CCNN. In this paper, we create a visual concept retrieval system which is based on CCNN. Experimental results on the CalTech101 dataset indicate that CCNN outperforms ANN.

In content based retrieval, color indexing is one of the most prevalent retrieval methods. The key problems in color indexing are: (1) the choice of color space, (2) color features, and (3) finding the best distance metric. In our color experiments we examine two applications from computer vision which involve distortions derived from changes in viewpoint and the process of printing and scanning. In the first experiments we use the Corel stock photo database and a color histogram method to find copies of images which were printed and subsequently scanned in. The second application deals with object based retrieval. The goal is to find all images of an object in a database where the images depicting the object were taken from different viewpoints. Both the ground truth and the algorithm come from the work by Gevers and Smeulders [Gevers and Smeulders, 1999]. Furthermore, for both applications, we implement the quadratic perceptual similarity measure proposed by Hafner et al. [Hafner et al., 1995] and the correlogram introduced by Huang et al. [Huang et al., 1997] as benchmarks.

Processing visual content in images and videos is a challenging task associated with the development of modern computer vision. Because salient point approaches can represent distinctive and affine invariant points in images, many approaches have been proposed over the past decade. Each method has particular advantages and limitations and may be appropriate in different contexts. In this paper we evaluate the performance of a wide set of salient point detectors and descriptors. We begin by comparing diverse salient point algorithms (SIFT, SURF, BRIEF, ORB, FREAK, BRISK, STAR, GFTT and FAST) with regard to repeatability, recall and precision and then move to accuracy and stability in real-time video tracking.

The world wide web is filled with billions of images and duplicates of images can frequently be found on many websites. These duplicates can be exact copies or differ slightly in their visual content. In this paper we provide a comparative study on how well content-based duplicate image detection methods are able to detect the duplicates of a query image. We conduct a survey to better understand in which ways such images on the internet differ from each other and use these observations to form a realistic and challenging duplicate image detection scenario. The methods we evaluate in our study are representative techniques from the research literature. In our evaluation, we target the performance of each method in relation to their descriptor size, description time and matching time, to assess their feasibility of application to large image collections (> 1 million).

We present a novel stereo matching algorithm which integrates learning, feature selection, and surface reconstruction. First, a new instance based learning (IBL) algorithm is used to generate an approximation to the optimal feature set for matching. In addition, the importance of two separate kinds of knowledge, image dependent knowledge and image independent knowledge, is discussed. Second, we develop an adaptive method for refining the feature set. This adaptive method analyzes the feature error to locate areas of the image that would lead to false matches. Then these areas are used to guide the search through feature space towards maximizing the class separation distance between the correct match and the false matches. Third, we introduce a self-diagnostic method for determining when apriori knowledge is necessary for finding the correct match. If the a priori knowledge is necessary then we use a surface reconstruction model to discriminate between match possibilities. Our algorithm is comprehensively tested against fixed feature set algorithms and against a traditional pyramid algorithm. Finally, we present and discuss extensive empirical results of our algorithm based on a large set of real images. >

In this paper we present RetrievalLab, a content based retrieval tool that was designed for both educational and research purposes. It is a tool to facilitate the testing of new features, segmentations, machine learning approaches, and evaluation methods, by presenting a Matlab-like programming interface which illuminates the fundamental processes and algorithms in content based retrieval.

Social computing examines the interactions between users and media across diverse cultures. Currently the Internet is the most prominent example and has fostered the development of new social interaction paradigms. This paper surveys the research literature in social computing and connected media and then introduces an emerging area, social webconsciousness, which describes the experiential nature of social interactions over the Internet.

TOP-SURF is an image descriptor that combines interest points with visual words, resulting in a high performance yet compact descriptor that is designed with a wide range of content-based image retrieval applications in mind. TOP-SURF offers the flexibility to vary descriptor size and supports very fast image matching. In addition to the source code for the visual word extraction and comparisons, we also provide a high level API and very large pre-computed codebooks targeting web image content for both research and teaching purposes.

Visual concept detection in images has been a challenging task for many years. The recently proposed MIRFLICKR-25000 dataset has set the standards even higher as the wide variety of images and annotations require new techniques to tackle the visual concept detection problem. We propose the use of the recently introduced MOD salient points for subimage visual concept detection. These points are located at regions within an image that are distinctive with respect to the features that are selected for subimage classification. We also introduce the notion of Minimum Explanation Complexity (MEC), where the complexity of classifiers is reduced to a simpler but equally effective form whenever possible. Our experiments on the MIRFLICKR-25000 dataset show that MOD based concept detectors outperform SIFT based features. We also show that a neural network classifier based on the MEC notion, outperforms a standard SVM classifier.

The MIR Flickr collection consists of 25000 high-quality photographic images of thousands of Flickr users, made available under the Creative Commons license. The database includes all the original user tags and EXIF metadata. Additionally, detailed and accurate annotations are provided for topics corresponding to the most prominent visual concepts in the user tag data. The rich metadata allow for a wide variety of image retrieval benchmarking scenarios.In this paper, we provide an overview of the various strategies that were devised for automatic visual concept detection using the MIR Flickr collection. In particular we discuss results from various experiments in combining social data and low-level content-based descriptors to improve the accuracy of visual concept classifiers. Additionally, we present retrieval results obtained by relevance feedback methods, demonstrating (i) how their performance can be enhanced using features based on visual concept classifiers, and (ii) how their performance, based on small samples, can be measured relative to their large sample classifier counterparts.Additionally, we identify a number of promising trends and ideas in visual concept detection. To keep the MIR Flickr collection up-to-date on these developments, we have formulated two new initiatives to extend the original image collection. First, the collection will be extended to one million Creative Commons Flickr images. Second, a number of state-of-the-art content-based descriptors will be made available for the entire collection.

Experiential image retrieval systems aim to provide the user with a natural and intuitive search experience. The goal is to empower the user to navigate large collections based on his own needs and preferences, while simultaneously providing him with an accurate sense of what the database has to offer. In this paper we integrate a new browsing mechanism called deep exploration with the proven technique of retrieval by relevance feedback. In our approach, relevance feedback focuses the search on relevant regions, while deep exploration facilitates transparent navigation to promising regions of feature space that would normally remain unreachable. Optimal feature weights are determined automatically based on the evidential support for the relevance of each single feature. To achieve efficient refinement of the search space, images are ranked and presented to the user based on their likelihood of being useful for further exploration.

One of the grand challenges in our field is considered to be the need for experiential exploration systems that allow the user to gain insight into and support exploration of media collections. In this paper we propose such a system, where a novel interface gives the user the opportunity to visually and interactively explore the feature space around relevant images and to focus the search on only those regions in feature space that are relevant. When an image is explored, its optimal set of feature weights is automatically determined using all images contained within the relevant regions, based on the evidential support for the relevance of each single feature. Images are ranked twofold, where one ranking reflects the likelihood the image is useful for further exploration of the feature space and the other reflects the likelihood the image is relevant to the user.

One well-known and effective method used for computationally efficient texture classification is the use of statistical information on 3×3 pixel blocks such as local binary patterns (LBP). However, there has been negligible research on sizes of pixel blocks beyond 3×3 while using the histogram approach. Specifically, larger or non-square features might give better classification results. Our proposed method constructs features, with arbitrary size and shape, that will give the best results for classifying a specific texture, and still keeping the feature vectors as short as possible. In this work, we show the selected features and the performance of our method with a minimum distance classifier and with a neural network and provide quantitative comparisons to the 3×3 block method on both the Brodatz and Ponce texture databases.

Our goal is to determine if artificially imagined or synthesized images can be beneficial to interactive visual search. We present a novel approach for using artificially imagined images in relevance feedback. Since the search engine constructs the synthetic images itself, any feedback given by the user on these images allows it to obtain a better understanding of what the user is looking for than it would from feedback on database images alone. We evaluated and compared our image synthesis approach with a normal Rocchio-based system on a well-known texture database with real users.

In this paper we provide a comparative study of content-based copy detection methods, which include research literature methods based on salient point matching (SURF), discrete cosine and wavelet transforms, color histograms, biologically motivated visual matching and other methods. In our evaluation we focus on large-scale applications, especially on performance in the context of search engines for web images. We assess the scalability of the tested methods by investigating the detection accuracy relative to descriptor size, description time per image and matching time per image. For testing, original images altered by a diverse set of realistic transformations are embedded in a collection of one million web images.

Interest or salient points are typically meaningful points within an image which can be used for a wide variety of image understanding tasks. In this paper we present a novel algorithm for detecting interest points within images. The new technique is based on finding the locations in an image which exhibit local distinctiveness. We evaluate our algorithm on the Corel stock photography test set in the context of content based image retrieval from large databases and provide quantitative comparisons to the well known SIFT interest point and Harris corner detectors as a benchmark.