Abstract:This paper aims to delve into the rate-distortion-complexity trade-offs of modern neural video coding. Recent years have witnessed much research effort being focused on exploring the full potential of neural video coding. Conditional autoencoders have emerged as the mainstream approach to efficient neural video coding. The central theme of conditional autoencoders is to leverage both spatial and temporal information for better conditional coding. However, a recent study indicates that conditional coding may suffer from information bottlenecks, potentially performing worse than traditional residual coding. To address this issue, recent conditional coding methods incorporate a large number of high-resolution features as the condition signal, leading to a considerable increase in the number of multiply-accumulate operations, memory footprint, and model size. Taking DCVC as the common code base, we investigate how the newly proposed conditional residual coding, an emerging new school of thought, and its variants may strike a better balance among rate, distortion, and complexity.
Abstract:Conditional coding has lately emerged as the mainstream approach to learned video compression. However, a recent study shows that it may perform worse than residual coding when the information bottleneck arises. Conditional residual coding was thus proposed, creating a new school of thought to improve on conditional coding. Notably, conditional residual coding relies heavily on the assumption that the residual frame has a lower entropy rate than that of the intra frame. Recognizing that this assumption is not always true due to dis-occlusion phenomena or unreliable motion estimates, we propose a masked conditional residual coding scheme. It learns a soft mask to form a hybrid of conditional coding and conditional residual coding in a pixel adaptive manner. We introduce a Transformer-based conditional autoencoder. Several strategies are investigated with regard to how to condition a Transformer-based autoencoder for inter-frame coding, a topic that is largely under-explored. Additionally, we propose a channel transform module (CTM) to decorrelate the image latents along the channel dimension, with the aim of using the simple hyperprior to approach similar compression performance to the channel-wise autoregressive model. Experimental results confirm the superiority of our masked conditional residual transformer (termed MaskCRT) to both conditional coding and conditional residual coding. On commonly used datasets, MaskCRT shows comparable BD-rate results to VTM-17.0 under the low delay P configuration in terms of PSNR-RGB. It also opens up a new research direction for advancing learned video compression.
Abstract:Aerial image analysis, specifically the semantic segmentation thereof, is the basis for applications such as automatically creating and updating maps, tracking city growth, or tracking deforestation. In true orthophotos, which are often used in these applications, many objects and regions can be approximated well by polygons. However, this fact is rarely exploited by state-of-the-art semantic segmentation models. Instead, most models allow unnecessary degrees of freedom in their predictions by allowing arbitrary region shapes. We therefore present a refinement of our deep learning model which predicts binary space partitioning trees, an efficient polygon representation. The refinements include a new feature decoder architecture and a new differentiable BSP tree renderer which both avoid vanishing gradients. Additionally, we designed a novel loss function specifically designed to improve the spatial partitioning defined by the predicted trees. Furthermore, our expanded model can predict multiple trees at once and thus can predict class-specific segmentations. Taking all modifications together, our model achieves state-of-the-art performance while using up to 60% fewer model parameters when using a small backbone model or up to 20% fewer model parameters when using a large backbone model.
Abstract:Human affect recognition is an essential part of natural human-computer interaction. However, current methods are still in their infancy, especially for in-the-wild data. In this work, we introduce our submission to the Affective Behavior Analysis in-the-wild (ABAW) 2020 competition. We propose a two-stream aural-visual analysis model to recognize affective behavior from videos. Audio and image streams are first processed separately and fed into a convolutional neural network. Instead of applying recurrent architectures for temporal analysis we only use temporal convolutions. Furthermore, the model is given access to additional features extracted during face-alignment. At training time, we exploit correlations between different emotion representations to improve performance. Our model achieves promising results on the challenging Aff-Wild2 database.
Abstract:The efficiency of motion compensated prediction in modern video codecs highly depends on the available reference pictures. Occlusions and non-linear motion pose challenges for the motion compensation and often result in high bit rates for the prediction error. We propose the generation of artificial reference pictures using deep recurrent neural networks. Conceptually, a reference picture at the time instance of the currently coded picture is generated from previously reconstructed conventional reference pictures. Based on these artificial reference pictures, we propose a complete coding pipeline based on HEVC. By using the artificial reference pictures for motion compensated prediction, average BD-rate gains of 1.5% over HEVC are achieved.
Abstract:With the deployment of neural networks on mobile devices and the necessity of transmitting neural networks over limited or expensive channels, the file size of the trained model was identified as bottleneck. In this paper, we propose a codec for the compression of neural networks which is based on transform coding for convolutional and dense layers and on clustering for biases and normalizations. By using this codec, we achieve average compression factors between 7.9-9.3 while the accuracy of the compressed networks for image classification decreases only by 1%-2%, respectively.
Abstract:The diversity of facial shapes and motions among persons is one of the greatest challenges for automatic analysis of facial expressions. In this paper, we propose a feature describing expression intensity over time, while being invariant to person and the type of performed expression. Our feature is a weighted combination of the dynamics of multiple points adapted to the overall expression trajectory. We evaluate our method on several tasks all related to temporal analysis of facial expression. The proposed feature is compared to a state-of-the-art method for expression intensity estimation, which it outperforms. We use our proposed feature to temporally align multiple sequences of recorded 3D facial expressions. Furthermore, we show how our feature can be used to reveal person-specific differences in performances of facial expressions. Additionally, we apply our feature to identify the local changes in face video sequences based on action unit labels. For all the experiments our feature proves to be robust against noise and outliers, making it applicable to a variety of applications for analysis of facial movements.