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

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

There is large consent that successful training of deep networks requires many thousand annotated training samples. In this paper, we present a network and training strategy that relies on the strong use of data augmentation to use the available annotated samples more efficiently. The architecture consists of a contracting path to capture context and a symmetric expanding path that enables precise localization. We show that such a network can be trained end-to-end from very few images and outperforms the prior best method (a sliding-window convolutional network) on the ISBI challenge for segmentation of neuronal structures in electron microscopic stacks. Using the same network trained on transmitted light microscopy images (phase contrast and DIC) we won the ISBI cell tracking challenge 2015 in these categories by a large margin. Moreover, the network is fast. Segmentation of a 512x512 image takes less than a second on a recent GPU. The full implementation (based on Caffe) and the trained networks are available at http://lmb.informatik.uni-freiburg.de/people/ronneber/u-net .

/pdf/u-net-convolutional-networks-for-biomedical-image-2xoyjzggxv.pdf

U-Net: Convolutional Networks for Biomedical Image Segmentation

We describe a two-stage method for detecting hands and their orientation in unconstrained images. The first stage uses three complementary detectors to propose hand bounding boxes. Each bounding box is then scored by the three detectors independently, and a second stage classifier learnt to compute a final confidence score for the proposals using these features. We make the following contributions: (i) we add context-based and skin-based proposals to a sliding window shape based detector to increase recall; (ii) we develop a new method of non-maximum suppression based on super-pixels; and (iii) we introduce a fully annotated hand dataset for training and testing. We show that the hand detector exceeds the state of the art on two public datasets, including the PASCAL VOC 2010 human layout challenge.

/pdf/hand-detection-using-multiple-proposals-3w7uy2uf7a.pdf

Hand detection using multiple proposals.

A structure from motion algorithm is described which recovers structure and camera position, modulo a projective ambiguity. Camera calibration is not required, and camera parameters such as focal length can be altered freely during motion. The structure is updated sequentially over an image sequence, in contrast to schemes which employ a batch process. A specialisation of the algorithm to recover structure and camera position modulo an affine transformation is described, together with a method to periodically update the affine coordinate frame to prevent drift over time. We describe the constraint used to obtain this specialisation.

Structure is recovered from image corners detected and matched automatically and reliably in real image sequences. Results are shown for reference objects and indoor environments, and accuracy of recovered structure is fully evaluated and compared for a number of reconstruction schemes. A specific application of the work is demonstrated—affine structure is used to compute free space maps enabling navigation through unstructured environments and avoidance of obstacles. The path planning involves only affine constructions.

Sequential Updating of Projective and Affine Structure from Motion

We describe the 2020 edition of the DeepMind Kinetics human action dataset, which replenishes and extends the Kinetics-700 dataset. In this new version, there are at least 700 video clips from different YouTube videos for each of the 700 classes. This paper details the changes introduced for this new release of the dataset and includes a comprehensive set of statistics as well as baseline results using the I3D network.

A Short Note on the Kinetics-700 Human Action Dataset

As virtual worlds demand ever more realistic 3D models, attention is being focussed on systems that can acquire graphical models from real objects. This paper describes a system which, given a sequence of images of an object rotating about a single axis, generates a textured 3D model fully automatically. In contrast to previous approaches, the technique described here requires no prior information about the cameras or scene, and does not require that the turntable angles be known (or even constant through the sequence).

From an analysis ofthe projective geometry of the situation, it is shown that the rotation angles may be determined unambiguously, and that camera calibration, camera positions and 3D structure may be determined to within a two parameter family. An algorithm has been implemented to compute this reconstruction fully automatically. The two parameter reconstruction ambiguity may be removed by specifying, for example, camera aspect ratio and parallel scene lines. Examples are presented on four turn-table sequences.

Andrew Zisserman

Papers

Hand detection using multiple proposals.

Sequential Updating of Projective and Affine Structure from Motion

A Short Note on the Kinetics-700 Human Action Dataset

Automatic 3D Model Construction for Turn-Table Sequences

Robust parameterization and computation of the trifocal tensor