The success of CNNs in various applications is accompanied by a significant increase in the computation and parameter storage costs. Recent efforts toward reducing these overheads involve pruning and compressing the weights of various layers without hurting original accuracy. However, magnitude-based pruning of weights reduces a significant number of parameters from the fully connected layers and may not adequately reduce the computation costs in the convolutional layers due to irregular sparsity in the pruned networks. We present an acceleration method for CNNs, where we prune filters from CNNs that are identified as having a small effect on the output accuracy. By removing whole filters in the network together with their connecting feature maps, the computation costs are reduced significantly. In contrast to pruning weights, this approach does not result in sparse connectivity patterns. Hence, it does not need the support of sparse convolution libraries and can work with existing efficient BLAS libraries for dense matrix multiplications. We show that even simple filter pruning techniques can reduce inference costs for VGG-16 by up to 34% and ResNet-110 by up to 38% on CIFAR10 while regaining close to the original accuracy by retraining the networks.

No abstract

This paper concentrates o n the visual front end for hidden Markov model based automatic lipreading. Two approaches for extracting features relevant to lipreading, given image sequences of the speaker's mouth region, are considered: A lip contour based feature approach, which first obtains estimates of the speaker's lip contours and subsequently extracts features from them, and an image transf o r m based approach, which obtains a compressed representation of the image pixel values that contain the speaker's mouth. Various possible features are considered i n each approach, and experimental results o n a number of visualonly recognition tasks are reported. It is shown that the image transform based approach results i n superior lipreading performance. In addition, feature mean subtraction is demonstrated to improve performance i n multi-speaker and speaker-independent recognition tasks. Finally, the effects of video degradations to image transform based automatic lipreading are studied. It is shown that lipreading performance dramatically deteriorates below a 10 Hz field rate, and that image transform features are robust to noise and compression artifacts.

No abstract

Human actions often involve complex interactions across several inter-related objects in the scene. However, existing approaches to fine-grained video understanding or visual relationship detection often rely on single object representation or pairwise object relationships. Furthermore, learning interactions across multiple objects in hundreds of frames for video is computationally infeasible and performance may suffer since a large combinatorial space has to be modeled. In this paper, we propose to efficiently learn higher-order interactions between arbitrary subgroups of objects for fine-grained video understanding. We demonstrate that modeling object interactions significantly improves accuracy for both action recognition and video captioning, while saving more than 3-times the computation over traditional pairwise relationships. The proposed method is validated on two large-scale datasets: Kinetics and ActivityNet Captions. Our SINet and SINet-Caption achieve state-of-the-art performances on both datasets even though the videos are sampled at a maximum of 1 FPS. To the best of our knowledge, this is the first work modeling object interactions on open domain large-scale video datasets, and we additionally model higher-order object interactions which improves the performance with low computational costs. * Work performed as a NEC Labs intern Attended interactions Interactions between ROIs Video frame Action prediction: cooking on campfire , cooking egg , …

No abstract

As we articulate speech, we usually move the head and exhibit various facial expressions. This visual aspect of speech aids understanding and helps communicating additional information, such as the speaker's mood. In this paper we analyze quantitatively head and facial movements that accompany speech and investigate how they relate to the text's prosodic structure. We recorded several hours of speech and measured the locations of the speakers' main facial features as well as their head poses. The text was evaluated with a prosody prediction tool, identifying phrase boundaries and pitch accents. Characteristic for most speakers are simple motion patterns that are repeatedly applied in synchrony with the main prosodic events. Direction and strength of head movements vary widely from one speaker to another, yet their timing is typically well synchronized with the spoken text. Understanding quantitatively the correlations between head movements and spoken text is important for synthesizing photo-realistic talking heads. Talking heads appear much more engaging when they exhibit realistic motion patterns.

Developing conditional generative models for text-to-video synthesis is an extremely challenging yet an important topic of research in machine learning. In this work, we address this problem by introducing Text-Filter conditioning Generative Adversarial Network (TFGAN), a conditional GAN model with a novel multi-scale text-conditioning scheme that improves text-video associations. By combining the proposed conditioning scheme with a deep GAN architecture, TFGAN generates high quality videos from text on challenging real-world video datasets. In addition, we construct a synthetic dataset of text-conditioned moving shapes to systematically evaluate our conditioning scheme. Extensive experiments demonstrate that TFGAN significantly outperforms existing approaches, and can also generate videos of novel categories not seen during training.