Over the last years deep learning methods have been shown to outperform previous state-of-the-art machine learning techniques in several fields, with computer vision being one of the most prominent cases. This review paper provides a brief overview of some of the most significant deep learning schemes used in computer vision problems, that is, Convolutional Neural Networks, Deep Boltzmann Machines and Deep Belief Networks, and Stacked Denoising Autoencoders. A brief account of their history, structure, advantages, and limitations is given, followed by a description of their applications in various computer vision tasks, such as object detection, face recognition, action and activity recognition, and human pose estimation. Finally, a brief overview is given of future directions in designing deep learning schemes for computer vision problems and the challenges involved therein.

/pdf/deep-learning-for-computer-vision-a-brief-review-31uavl0qvq.pdf

Deep Learning for Computer Vision: A Brief Review.

The purpose of this study is to determine whether current video datasets have sufficient data for training very deep convolutional neural networks (CNNs) with spatio-temporal three-dimensional (3D) kernels. Recently, the performance levels of 3D CNNs in the field of action recognition have improved significantly. However, to date, conventional research has only explored relatively shallow 3D architectures. We examine the architectures of various 3D CNNs from relatively shallow to very deep ones on current video datasets. Based on the results of those experiments, the following conclusions could be obtained: (i) ResNet-18 training resulted in significant overfitting for UCF-101, HMDB-51, and ActivityNet but not for Kinetics. (ii) The Kinetics dataset has sufficient data for training of deep 3D CNNs, and enables training of up to 152 ResNets layers, interestingly similar to 2D ResNets on ImageNet. ResNeXt-101 achieved 78.4% average accuracy on the Kinetics test set. (iii) Kinetics pretrained simple 3D architectures outperforms complex 2D architectures, and the pretrained ResNeXt-101 achieved 94.5% and 70.2% on UCF-101 and HMDB-51, respectively. The use of 2D CNNs trained on ImageNet has produced significant progress in various tasks in image. We believe that using deep 3D CNNs together with Kinetics will retrace the successful history of 2D CNNs and ImageNet, and stimulate advances in computer vision for videos. The codes and pretrained models used in this study are publicly available1.

/pdf/can-spatiotemporal-3d-cnns-retrace-the-history-of-2d-cnns-4sp7xydev9.pdf

Can Spatiotemporal 3D CNNs Retrace the History of 2D CNNs and ImageNet

With recent progress in graphics, it has become more tractable to train models on synthetic images, potentially avoiding the need for expensive annotations. However, learning from synthetic images may not achieve the desired performance due to a gap between synthetic and real image distributions. To reduce this gap, we propose Simulated+Unsupervised (S+U) learning, where the task is to learn a model to improve the realism of a simulators output using unlabeled real data, while preserving the annotation information from the simulator. We develop a method for S+U learning that uses an adversarial network similar to Generative Adversarial Networks (GANs), but with synthetic images as inputs instead of random vectors. We make several key modifications to the standard GAN algorithm to preserve annotations, avoid artifacts, and stabilize training: (i) a self-regularization term, (ii) a local adversarial loss, and (iii) updating the discriminator using a history of refined images. We show that this enables generation of highly realistic images, which we demonstrate both qualitatively and with a user study. We quantitatively evaluate the generated images by training models for gaze estimation and hand pose estimation. We show a significant improvement over using synthetic images, and achieve state-of-the-art results on the MPIIGaze dataset without any labeled real data.

/pdf/learning-from-simulated-and-unsupervised-images-through-uh69fs5u48.pdf

Learning from Simulated and Unsupervised Images through Adversarial Training

The Mapillary Vistas Dataset is a novel, large-scale street-level image dataset containing 25000 high-resolution images annotated into 66 object categories with additional, instance-specific labels for 37 classes. Annotation is performed in a dense and fine-grained style by using polygons for delineating individual objects. Our dataset is 5× larger than the total amount of fine annotations for Cityscapes and contains images from all around the world, captured at various conditions regarding weather, season and daytime. Images come from different imaging devices (mobile phones, tablets, action cameras, professional capturing rigs) and differently experienced photographers. In such a way, our dataset has been designed and compiled to cover diversity, richness of detail and geographic extent. As default benchmark tasks, we define semantic image segmentation and instance-specific image segmentation, aiming to significantly further the development of state-of-the-art methods for visual road-scene understanding.

The Mapillary Vistas Dataset for Semantic Understanding of Street Scenes

Datasets drive vision progress, yet existing driving datasets are impoverished in terms of visual content and supported tasks to study multitask learning for autonomous driving. Researchers are usually constrained to study a small set of problems on one dataset, while real-world computer vision applications require performing tasks of various complexities. We construct BDD100K, the largest driving video dataset with 100K videos and 10 tasks to evaluate the exciting progress of image recognition algorithms on autonomous driving. The dataset possesses geographic, environmental, and weather diversity, which is useful for training models that are less likely to be surprised by new conditions. Based on this diverse dataset, we build a benchmark for heterogeneous multitask learning and study how to solve the tasks together. Our experiments show that special training strategies are needed for existing models to perform such heterogeneous tasks. BDD100K opens the door for future studies in this important venue.

/pdf/bdd100k-a-diverse-driving-dataset-for-heterogeneous-1v7dodhb6h.pdf

BDD100K: A Diverse Driving Dataset for Heterogeneous Multitask Learning

Many recent advancements in Computer Vision are attributed to large datasets. Open-source software packages for Machine Learning and inexpensive commodity hardware have reduced the barrier of entry for exploring novel approaches at scale. It is possible to train models over millions of examples within a few days. Although large-scale datasets exist for image understanding, such as ImageNet, there are no comparable size video classification datasets. 
In this paper, we introduce YouTube-8M, the largest multi-label video classification dataset, composed of ~8 million videos (500K hours of video), annotated with a vocabulary of 4800 visual entities. To get the videos and their labels, we used a YouTube video annotation system, which labels videos with their main topics. While the labels are machine-generated, they have high-precision and are derived from a variety of human-based signals including metadata and query click signals. We filtered the video labels (Knowledge Graph entities) using both automated and manual curation strategies, including asking human raters if the labels are visually recognizable. Then, we decoded each video at one-frame-per-second, and used a Deep CNN pre-trained on ImageNet to extract the hidden representation immediately prior to the classification layer. Finally, we compressed the frame features and make both the features and video-level labels available for download. 
We trained various (modest) classification models on the dataset, evaluated them using popular evaluation metrics, and report them as baselines. Despite the size of the dataset, some of our models train to convergence in less than a day on a single machine using TensorFlow. We plan to release code for training a TensorFlow model and for computing metrics.

YouTube-8M: A Large-Scale Video Classification Benchmark

Predicting the future behavior of moving agents is essential for real world applications. It is challenging as the intent of the agent and the corresponding behavior is unknown and intrinsically multimodal. Our key insight is that for prediction within a moderate time horizon, the future modes can be effectively captured by a set of target states. This leads to our target-driven trajectory prediction (TNT) framework. TNT has three stages which are trained end-to-end. It first predicts an agent's potential target states $T$ steps into the future, by encoding its interactions with the environment and the other agents. TNT then generates trajectory state sequences conditioned on targets. A final stage estimates trajectory likelihoods and a final compact set of trajectory predictions is selected. This is in contrast to previous work which models agent intents as latent variables, and relies on test-time sampling to generate diverse trajectories. We benchmark TNT on trajectory prediction of vehicles and pedestrians, where we outperform state-of-the-art on Argoverse Forecasting, INTERACTION, Stanford Drone and an in-house Pedestrian-at-Intersection dataset.

TNT: Target-driveN Trajectory Prediction

https://www.microsoft.com/en-us/research/wp-content/uploads/2016/02/ActiveSemi-JCSL2009-inpress.pdf

Active learning and semi-supervised learning for speech recognition: A unified framework using the global entropy reduction maximization criterion

Accurate unsupervised learning of phonemes of a language directly from speech is demonstrated via an algorithm for joint unsupervised learning of the topology and parameters of a hidden Markov model (HMM); states and short state-sequences through this HMM correspond to the learnt sub-word units. The algorithm, originally proposed for unsupervised learning of allophonic variations within a given phoneme set, has been adapted to learn without any knowledge of the phonemes. An evaluation methodology is also proposed, whereby the state-sequence that aligns to a test utterance is transduced in an automatic manner to a phoneme-sequence and compared to its manual transcription. Over 85% phoneme recognition accuracy is demonstrated for speaker-dependent learning from fluent, large-vocabulary speech.

/pdf/unsupervised-learning-of-acoustic-sub-word-units-45eji6gr0b.pdf

Unsupervised Learning of Acoustic Sub-word Units

This paper addresses automatic skill assessment in robotic minimally invasive surgery. Hidden Markov models (HMMs) are developed for individual surgical gestures (or surgemes ) that comprise a typical bench-top surgical training task. It is known that such HMMs can be used to recognize and segment surgemes in previously unseen trials [1]. Here, the topology of each surgeme HMM is designed in a data-driven manner, mixing trials from multiple surgeons with varying skill levels, resulting in HMM states that model skill-specific sub-gestures . The sequence of HMM states visited while performing a surgeme are therefore indicative of the surgeon's skill level. This expectation is confirmed by the average edit distance between the state-level "transcripts" of the same surgeme performed by two surgeons with different expertise levels. Some surgemes are further shown to be more indicative of skill than others.

/pdf/data-derived-models-for-segmentation-with-application-to-5ep18ss8ej.pdf

Balakrishnan Varadarajan

Papers

YouTube-8M: A Large-Scale Video Classification Benchmark

TNT: Target-driveN Trajectory Prediction

Active learning and semi-supervised learning for speech recognition: A unified framework using the global entropy reduction maximization criterion

Unsupervised Learning of Acoustic Sub-word Units

Data-Derived Models for Segmentation with Application to Surgical Assessment and Training