Going deeper with convolutions
07 Jun 2015-pp 1-9
TL;DR: Inception as mentioned in this paper is a deep convolutional neural network architecture that achieves the new state of the art for classification and detection in the ImageNet Large-Scale Visual Recognition Challenge 2014 (ILSVRC14).
Abstract: We propose a deep convolutional neural network architecture codenamed Inception that achieves the new state of the art for classification and detection in the ImageNet Large-Scale Visual Recognition Challenge 2014 (ILSVRC14). The main hallmark of this architecture is the improved utilization of the computing resources inside the network. By a carefully crafted design, we increased the depth and width of the network while keeping the computational budget constant. To optimize quality, the architectural decisions were based on the Hebbian principle and the intuition of multi-scale processing. One particular incarnation used in our submission for ILSVRC14 is called GoogLeNet, a 22 layers deep network, the quality of which is assessed in the context of classification and detection.
Citations
More filters
27 Jun 2016
TL;DR: In this article, the authors proposed a residual learning framework to ease the training of networks that are substantially deeper than those used previously, which won the 1st place on the ILSVRC 2015 classification task.
Abstract: Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers—8× deeper than VGG nets [40] but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions1, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.
123,388 citations
Proceedings Article•
01 Jan 2015TL;DR: In this paper, the authors investigated the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting and showed that a significant improvement on the prior-art configurations can be achieved by pushing the depth to 16-19 layers.
Abstract: In this work we investigate the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting. Our main contribution is a thorough evaluation of networks of increasing depth using an architecture with very small (3x3) convolution filters, which shows that a significant improvement on the prior-art configurations can be achieved by pushing the depth to 16-19 weight layers. These findings were the basis of our ImageNet Challenge 2014 submission, where our team secured the first and the second places in the localisation and classification tracks respectively. We also show that our representations generalise well to other datasets, where they achieve state-of-the-art results. We have made our two best-performing ConvNet models publicly available to facilitate further research on the use of deep visual representations in computer vision.
49,914 citations
Posted Content•
TL;DR: This work presents a residual learning framework to ease the training of networks that are substantially deeper than those used previously, and provides comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth.
Abstract: Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers---8x deeper than VGG nets but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers.
The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.
44,703 citations
Book•
[...]
TL;DR: Deep learning as mentioned in this paper is a form of machine learning that enables computers to learn from experience and understand the world in terms of a hierarchy of concepts, and it is used in many applications such as natural language processing, speech recognition, computer vision, online recommendation systems, bioinformatics, and videogames.
Abstract: Deep learning is a form of machine learning that enables computers to learn from experience and understand the world in terms of a hierarchy of concepts. Because the computer gathers knowledge from experience, there is no need for a human computer operator to formally specify all the knowledge that the computer needs. The hierarchy of concepts allows the computer to learn complicated concepts by building them out of simpler ones; a graph of these hierarchies would be many layers deep. This book introduces a broad range of topics in deep learning. The text offers mathematical and conceptual background, covering relevant concepts in linear algebra, probability theory and information theory, numerical computation, and machine learning. It describes deep learning techniques used by practitioners in industry, including deep feedforward networks, regularization, optimization algorithms, convolutional networks, sequence modeling, and practical methodology; and it surveys such applications as natural language processing, speech recognition, computer vision, online recommendation systems, bioinformatics, and videogames. Finally, the book offers research perspectives, covering such theoretical topics as linear factor models, autoencoders, representation learning, structured probabilistic models, Monte Carlo methods, the partition function, approximate inference, and deep generative models. Deep Learning can be used by undergraduate or graduate students planning careers in either industry or research, and by software engineers who want to begin using deep learning in their products or platforms. A website offers supplementary material for both readers and instructors.
38,208 citations
TL;DR: A large, deep convolutional neural network was trained to classify the 1.2 million high-resolution images in the ImageNet LSVRC-2010 contest into the 1000 different classes and employed a recently developed regularization method called "dropout" that proved to be very effective.
Abstract: We trained a large, deep convolutional neural network to classify the 1.2 million high-resolution images in the ImageNet LSVRC-2010 contest into the 1000 different classes. On the test data, we achieved top-1 and top-5 error rates of 37.5% and 17.0%, respectively, which is considerably better than the previous state-of-the-art. The neural network, which has 60 million parameters and 650,000 neurons, consists of five convolutional layers, some of which are followed by max-pooling layers, and three fully connected layers with a final 1000-way softmax. To make training faster, we used non-saturating neurons and a very efficient GPU implementation of the convolution operation. To reduce overfitting in the fully connected layers we employed a recently developed regularization method called "dropout" that proved to be very effective. We also entered a variant of this model in the ILSVRC-2012 competition and achieved a winning top-5 test error rate of 15.3%, compared to 26.2% achieved by the second-best entry.
33,301 citations
References
More filters
07 Jun 2015
TL;DR: This paper proposes a joint reconstruction and retrieval system that maintains the scalability of large-scale Structure-from-Motion systems while also recovering the often lost ability of reconstructing fine details of the scene.
Abstract: Structure-from-Motion for unordered image collections has significantly advanced in scale over the last decade. This impressive progress can be in part attributed to the introduction of efficient retrieval methods for those systems. While this boosts scalability, it also limits the amount of detail that the large-scale reconstruction systems are able to produce. In this paper, we propose a joint reconstruction and retrieval system that maintains the scalability of large-scale Structure-from-Motion systems while also recovering the often lost ability of reconstructing fine details of the scene. We demonstrate our proposed method on a large-scale dataset of 7.4 million images downloaded from the Internet.
144 citations
16 Jun 2012
TL;DR: It is shown that image sequences with more frames are needed to resolve ambiguities in depth ordering at occlusion boundaries; temporal layer constancy makes this feasible and the optimizer, which mixes discrete and continuous optimization, automatically determines the number of layers and reasons about their depth ordering.
Abstract: Layered models provide a compelling approach for estimating image motion and segmenting moving scenes. Previous methods, however, have failed to capture the structure of complex scenes, provide precise object boundaries, effectively estimate the number of layers in a scene, or robustly determine the depth order of the layers. Furthermore, previous methods have focused on optical flow between pairs of frames rather than longer sequences. We show that image sequences with more frames are needed to resolve ambiguities in depth ordering at occlusion boundaries; temporal layer constancy makes this feasible. Our generative model of image sequences is rich but difficult to optimize with traditional gradient descent methods. We propose a novel discrete approximation of the continuous objective in terms of a sequence of depth-ordered MRFs and extend graph-cut optimization methods with new “moves” that make joint layer segmentation and motion estimation feasible. Our optimizer, which mixes discrete and continuous optimization, automatically determines the number of layers and reasons about their depth ordering. We demonstrate the value of layered models, our optimization strategy, and the use of more than two frames on both the Middlebury optical flow benchmark and the MIT layer segmentation benchmark.
132 citations
05 Nov 2012
TL;DR: The results show that a frame-based superpixel segmentation combined with a few motion and appearance-based affinities are sufficient to obtain good video segmentation performance.
Abstract: Due to its importance, video segmentation has regained interest recently. However, there is no common agreement about the necessary ingredients for best performance. This work contributes a thorough analysis of various within- and between-frame affinities suitable for video segmentation. Our results show that a frame-based superpixel segmentation combined with a few motion and appearance-based affinities are sufficient to obtain good video segmentation performance. A second contribution of the paper is the extension of [1] to include motion-cues, which makes the algorithm globally aware of motion, thus improving its performance for video sequences. Finally, we contribute an extension of an established image segmentation benchmark [1] to videos, allowing coarse-to-fine video segmentations and multiple human annotations. Our results are tested on BMDS [2], and compared to existing methods.
130 citations
01 Dec 2013
TL;DR: This paper will show how simple moment based algorithms - such as Gray-World - can, with the addition of a simple correction step, deliver much improved illuminant estimation performance.
Abstract: Image colors are biased by the color of the prevailing illumination. As such the color at pixel cannot always be used directly in solving vision tasks from recognition, to tracking to general scene understanding. Illuminant estimation algorithms attempt to infer the color of the light incident in a scene and then a color cast removal step discounts the color bias due to illumination. However, despite sustained research since almost the inception of computer vision, progress has been modest. The best algorithms - now often built on top of expensive feature extraction and machine learning - are only about twice as good as the simplest approaches. This paper, in effect, will show how simple moment based algorithms - such as Gray-World - can, with the addition of a simple correction step, deliver much improved illuminant estimation performance. The corrected Gray-World algorithm maps the mean image color using a fixed (per camera) 3x3 matrix transform. More generally, our moment approach employs 1st, 2nd and higher order moments - of colors or features such as color derivatives - and these again are linearly corrected to give an illuminant estimate. The question of how to correct the moments is an important one yet we will show a simple alternating least-squares training procedure suffices. Remarkably, across the major datasets - evaluated using a 3-fold cross validation procedure - our simple corrected moment approach always delivers the best results (and the performance increment is often large compared with the prior art). Significantly, outlier performance was found to be much improved.
127 citations
Proceedings Article•
28 Jun 2011TL;DR: This work addresses the problem of metric learning for multi-view data, namely the construction of embedding projections from data in different representations into a shared feature space, such that the Euclidean distance in this space provides a meaningful within-view as well as between-view similarity.
Abstract: We address the problem of metric learning for multi-view data, namely the construction of embedding projections from data in different representations into a shared feature space, such that the Euclidean distance in this space provides a meaningful within-view as well as between-view similarity. Our motivation stems from the problem of cross-media retrieval tasks, where the availability of a joint Euclidean distance function is a prerequisite to allow fast, in particular hashing-based, nearest neighbor queries.
We formulate an objective function that expresses the intuitive concept that matching samples are mapped closely together in the output space, whereas non-matching samples are pushed apart, no matter in which view they are available. The resulting optimization problem is not convex, but it can be decomposed explicitly into a convex and a concave part, thereby allowing efficient optimization using the convex-concave procedure. Experiments on an image retrieval task show that nearest-neighbor based cross-view retrieval is indeed possible, and the proposed technique improves the retrieval accuracy over baseline techniques.
125 citations