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.

/pdf/deep-residual-learning-for-image-recognition-b75jg61qrq.pdf

Deep Residual Learning for Image Recognition

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.

Very Deep Convolutional Networks for Large-Scale Image Recognition

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.

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.

/pdf/going-deeper-with-convolutions-1yobw2o2ds.pdf

Going deeper with convolutions

: This paper presents methods to visualize feature spaces commonly used in object detection. The tools in this paper allow a human to put on feature space glasses and see the visual world as a computer might see it. We found that these glasses allow us to gain insight into the behavior of computer vision systems. We show a variety of experiments with our visualizations, such as examining the linear separability of recognition in HOG space, generating high scoring super objects for an object detector, and diagnosing false positives. We pose the visualization problem as one of feature inversion, i.e. recovering the natural image that generated a feature descriptor. We describe four algorithms to tackle this task, with different trade-offs in speed accuracy, and scalability. Our most successful algorithm uses ideas from sparse coding to learn a pair of dictionaries that enable regression between HOG features and natural images, and can invert features at interactive rates. We believe these visualizations are useful tools to add to an object detector researcher's toolbox, and code is available.

Inverting and Visualizing Features for Object Detection

Scene understanding is the ability to visually analyze a scene to answer questions such as: What is happening? Why is it happening? What will happen next? What should I do? For example, in the context of driving safety, the vision system would need to recognize nearby people and vehicles, anticipate their motions, infer traffic patterns, and detect road conditions. So far, research has focused on providing complete (e.g., every pixel labeled) or holistic (reasoning about several different scene elements) interpretations, often taking into account scene geometry or 3D spatial relationships. Accordingly, in this issue, several papers offer improvements to image segmentation and labeling through use of region classifiers, detectors, and object and scene context: “Indoor Scene Understanding with RGB-D Images: Bottom-up Segmentation, Object Detection and Semantic Segmentation” (doi:10.1007/s11263-014-0777-6) by Gupta et al. addresses problems of interpreting indoor scenes from a paired RGB and depth image. The method infers whether observed contours are due to depth, normal, or albedo

/pdf/guest-editorial-scene-understanding-5a2jcgx6at.pdf

Guest Editorial: Scene Understanding

In this paper we consider the pollution attack in network coded systems where network nodes are computationally limited. We consider the combined use of cryptographic signature based security and information theoretic network error correction and propose a fountain-like network error correction code construction suitable for this purpose.

/pdf/on-combining-information-theoretic-and-cryptographic-uuyft49jtn.pdf

On combining information-theoretic and cryptographic approaches to network coding security against the pollution attack

We introduce algorithms to visualize feature spaces used by object detectors. Our method works by inverting a visual feature back to multiple natural images. We found that these visualizations allow us to analyze object detection systems in new ways and gain new insight into the detector's failures. For example, when we visualize the features for high scoring false alarms, we discovered that, although they are clearly wrong in image space, they do look deceptively similar to true positives in feature space. This result suggests that many of these false alarms are caused by our choice of feature space, and supports that creating a better learning algorithm or building bigger datasets is unlikely to correct these errors. By visualizing feature spaces, we can gain a more intuitive understanding of recognition systems.

Visualizing Object Detection Features

The neural machinery underlying visual object recognition comprises a hierarchy of cortical regions in the ventral visual stream. The spatiotemporal dynamics of information flow in this hierarchy of regions is largely unknown. Here we tested the hypothesis that there is a correspondence between the spatiotemporal neural processes in the human brain and the layer hierarchy of a deep convolutional neural network (CNN). We presented 118 images of real-world objects to human participants (N=15) while we measured their brain activity with functional magnetic resonance imaging (fMRI) and magnetoencephalography (MEG). We trained an 8 layer (5 convolutional layers, 3 fully connected layers) CNN to predict 683 object categories with 900K training images from the ImageNet dataset. We obtained layer-specific CNN responses to the same 118 images. To compare brain-imaging data with the CNN in a common framework, we used representational similarity analysis. The key idea is that if two conditions evoke similar patterns in brain imaging data, they should also evoke similar patterns in the computer model. We thus determined 'where' (fMRI) and 'when' (MEG) the CNNs predicted brain activity. We found a correspondence in hierarchy between cortical regions, processing time, and CNN layers. Low CNN layers predicted MEG activity early and high layers relatively later; low CNN layers predicted fMRI activity in early visual regions, and high layers in late visual regions. Surprisingly, the correspondence between CNN layer hierarchy and cortical regions held for the ventral and dorsal visual stream. Results were dependent on amount of training and type of training material. Our results show that CNNs are a promising formal model of human visual object recognition. Combined with fMRI and MEG, they provide an integrated spatiotemporal and algorithmically explicit view of the first few hundred milliseconds of object recognition. Meeting abstract presented at VSS 2015.

Aditya Khosla

Papers

Inverting and Visualizing Features for Object Detection

Guest Editorial: Scene Understanding

On combining information-theoretic and cryptographic approaches to network coding security against the pollution attack

Visualizing Object Detection Features

Mapping human visual representations in space and time by neural networks