A QP-adaptive Mechanism for CNN-based Filter in Video Coding

High-Efficiency Video Coding (HEVC) is the latest video coding standard developed by the Joint Collaborative Team on Video Coding (JCT-VC). An effective rate control mechanism plays a critical role in making the best possible use of the network resources. The coding performance can be maximised through the appropriate allocation of bits under the constraints of a total bit rate budget and the buffer size. Existing rate control strategies generally determine the rate control parameters according to previous encoding results. However, there is no priori-knowledge available when the first frame of a video sequence is encoded or scene changes occur. Consequently, it is challenging to choose appropriate rate control parameters in such cases. In order to address this issue, we investigated deep learning oriented methods and proposed a Convolutional Neural Network (CNN)-based rate control scheme for HEVC intra-coding. As the CNN has the advantage of local perception and the Coding Tree Unit's (CTU's) Quantisation Parameter (QP) value is determined during the CTU level rate control process, we applied the CNN at the CTU level as well to achieve an accurate QP prediction. An improved QP determination model (R-QP model) involving only two model coefficients (µ and v) was suggested to simplify the QP calculation process. A CNN was used to predict the model coefficients, namely µ and v. Our CNN has four convolution layers, two pooling layers and three fully connected layers. Each of the convolution layers contains a Rectified Linear Unit (ReLU). The last fully connected layer outputs the predicted values of µ and v. The images from the UCID dataset and the RAISE dataset were used to train the CNN. In order to obtain a universal model, none of the frames in the HEVC common test sequences were employed as training images. In order to achieve a reasonable bit budget allocation, a CNN-based framework was used to predict the bit consumption for each CTU according to the entire picture content that it contains. At the training stage, the network took the original images and the QP values as inputs, and the corresponding actual bit consumptions were used as the labels. Once the CNN coefficients have been obtained through training, the bit consumption for each CTU under different QP values can be directly predicted from the original images. The predicted bit consumption was subsequently used to readjust the bit budget for each CTU to achieve a more reasonable bit allocation. We implemented the proposed CNN-based rate control scheme in the HEVC reference software HM 16.9. As the priori-knowledge of the previous encoding results had not been considered in the proposed algorithm, only the first frame of each video sequence was encoded. The experimental results showed that our proposed scheme performs better than the default rate control implementation in the HEVC reference software, namely the BDBR is reduced by 1.33% on average while maintaining the same reconstructed picture quality. Compared with the state-of-the-art CNN-based rate control algorithm proposed by Li, our schemes achieved better coding efficiency. It can be concluded that the proposed R-QP model is effective in describing the rate-distortion relationship in HEVC, and the CNN-based bit allocation scheme is appropriate for ensuring that the best use of each bit is made.

Convolutional Neural Network (CNN) is a powerful and successful deep learning technique for a variety of computer vision and image analysis applications. Interpreting and explaining the decisions of CNNs is one of the most challengeable tasks despite its significant success in various image analysis tasks. Topological Data Analysis (TDA) is an approach that exploits algebraic invariants from topology to analyse high dimensional and noisy datasets as well as growing challenges of big data applications. Persistent homology (PH) is an algebraic topology method for measuring topological features of shapes and/or functions at different distance or similarity resolutions. This work is an attempt to investigate the algebraic properties of pretrained CNN convolutional layer filters based on random Gaussian/Uniform distribution. We shall investigate the stability and sensitivity of the condition number of CNN filters during and post the model training with focus on class discriminability of the PH features of the convolved images. We shall demonstrate a strong link between the condition number of the CNN filters and their discriminating power of the PH representation. In particular, we shall establish that if small perturbation added to the original images then feature maps with well-conditioned filters will produce similar topological features to the original image. Our investigation and findings are based on training CNN's with Digits, MNIST and CIFAR-10 datasets. Our ultimate interest in applying the results of these findings in designing appropriate CNN models for classifications of ultrasound tumor scan images. Preliminary results for these applications are encouraging.

Symmetry is present in nature and science. In image processing, kernels for spatial filtering possess some symmetry (e.g. Sobel operators, Gaussian, Laplacian). Convolutional layers in artificial feed-forward neural networks have typically considered the kernel weights without any constraint. We propose to investigate the impact of a symmetry constraint in convolutional layers for image classification tasks, taking our inspiration from the processes involved in the primary visual cortex and common image processing techniques. The goal is to determine if it is necessary to learn each weight of the filters independently, and the extent to which it is possible to enforce symmetrical constraints on the filters throughout the training process of a convolutional neural network by modifying the weight update preformed during the backpropagation algorithm and to evaluate the change in performance. The symmetrical constraint reduces the number of free parameters in the network, and it is able to achieve near identical performance. We address the following cases: x/y-axis symmetry, point reflection, and anti-point reflection. The performance is evaluated on four databases of images representing handwritten digits. The results support the conclusion that while random weights offer more freedom to the model, the symmetry constraint provides a similar level of performance while decreasing substantially the number of free parameters in the model. Such an approach can be valuable in phase-sensitive applications that require a linear phase property throughout the feature extraction process.

As the upcoming video coding standard, Versatile Video Coding (i.e., VVC) achieves up to 30% Bjøntegaard delta bit-rate (BD-rate) reduction compared with High Efficiency Video Coding (H.265/HEVC). To eliminate or alleviate different kinds of compression artifacts like blocking, ringing, blurring and contouring effects, three in-loop filters, i.e. de-blocking filter (DBF), sample adaptive offset (SAO) and adaptive loop filter (ALF), have been involved in VVC. Recently, Convolutional Neural Network (CNN) has attracted tremendous attention and shows great potential in many tasks in image processing. In this work, we design a CNN-based in-loop filter as an integrated single-model solution which is adaptive to almost any scenarios in video coding. An architecture named as ADCNN (i.e., Attention based Dual-scale CNN) with an attention based processing block is proposed to reduce artifacts of I frames and B frames, which take advantage of informative priors such as the quantization parameter (QP) and partitioning information. Different from existing CNN-based filtering methods, which are mainly designed for the luma component and may need to train different models for different QPs, the proposed filter is adapted to different QPs and different frame types, and all the components (i.e., both luma and chroma) are processed simultaneously with feature exchange and fusion between components for information supplementary. Experimental results show that the proposed ADCNN filter can achieve 6.54%, 13.27%, 15.72% BD-rate savings for Y, U, V respectively under the all intra configuration and 2.81%, 7.86%, 8.60% BD-rate savings under the random access configuration. It can be used to replace all the conventional in-loop filters and also outperforms them without increase in encoding time.

High resolution (HR) videos are easy to go beyond the band-width limit during transmission after encoding. Downsampling followed by upsampling is a well-known strategy for compressing HR videos with limited bandwidth. Versatile video coding (VVC) provides a reference picture resampling (RPR) functionality to consider the limited bandwidth. In this paper, we propose a convolutional neural network (CNN) filter for super-resolution (SR) with the RPR functionality in VVC. We design a lightweight SR network that combines CNN with the RPR functionality in VVC, called lightweight network of multi-level mixed scale and depth information with attention mechanism (LMSDANet). For LMSDANet, we provide a lightweight block of multi-mixed scale and depth information with attention (LMSDAB) to extract multi-scale and convolutional layer depth information while enhancing the representation of features. Compared with VTM-11.0_NNVC-2.0 anchor, LMSDANet achieves {-9.16% (Y), 17.03% (U), -7.61% (V)} and {-4.14% (Y), 6.34% (U), -2.25% (V)} BD-rate changes (average on A1 and A2) in AI and RA configurations, respectively.

The concept of conditional computation for deep nets has been proposed previously to improve model performance by selectively using only parts of the model conditioned on the sample it is processing. In this paper, we investigate input-dependent dynamic filter selection in deep convolutional neural networks (CNNs). The problem is interesting because the idea of forcing different parts of the model to learn from different types of samples may help us acquire better filters in CNNs, improve the model generalization performance and potentially increase the interpretability of model behavior. We propose a novel yet simple framework called GaterNet, which involves a backbone and a gater network. The backbone network is a regular CNN that performs the major computation needed for making a prediction, while a global gater network is introduced to generate binary gates for selectively activating filters in the backbone network based on each input. Extensive experiments on CIFAR and ImageNet datasets show that our models consistently outperform the original models with a large margin. On CIFAR-10, our model also improves upon state-of-the-art results. © 2019 IEEE.

IIRI-Net: An interpretable convolutional front-end inspired by IIR filters for speaker identification

Suppression of interference fringes in wavelength modulation spectroscopy based on convolutional neural networks

The application of deep learning technology has increased rapidly in recent years. Technologies in deep learning increasingly emulate natural human abilities, such as knowledge learning, problem-solving, and decision-making. In general, deep learning can carry out self-training without repetitive programming by humans. Convolutional neural networks (CNNs) are deep learning algorithms commonly used in wide applications. CNN is often used for image classification, segmentation, object detection, video processing, natural language processing, and speech recognition. CNN has four layers: convolution layer, pooling layer, fully connected layer, and non-linear layer. The convolutional layer uses kernel filters to calculate the convolution of the input image by extracting the fundamental features. The pooling layer combines two successive convolutional layers. The third layer is the fully connected layer, commonly called the convolutional output layer. The activation function defines the output of a neural network, such as 'yes' or 'no'. The most common and popular CNN activation functions are Sigmoid, Tanh, ReLU, Leaky ReLU, Noisy ReLU, and Parametric Linear Units. The organization and function of the visual cortex greatly influence CNN architecture because it is designed to resemble the neuronal connections in the human brain. Some of the popular CNN architectures are LeNet, AlexNet and VGGNet.

Filter pruning is a technique that reduces computational complexity, inference time, and memory footprint by removing unnecessary filters in convolutional neural networks (CNNs) with an acceptable drop in accuracy, consequently accelerating the network. Unlike traditional filter pruning methods utilizing zeroing-out filters, we propose two techniques to achieve the effect of pruning more filters with less performance degradation, inspired by the existing research on centripetal stochastic gradient descent (C-SGD), wherein the filters are removed only when the ones that need to be pruned have the same value. First, to minimize the negative effect of centripetal vectors that gradually make filters come closer to each other, we redesign the vectors by considering the effect of each vector on the loss-function using the Taylor-based method. Second, we propose an adaptive gradient learning (AGL) technique that updates weights while adaptively changing the gradients. Through AGL, performance degradation can be mitigated because some gradients maintain their original direction, and AGL also minimizes the accuracy loss by perfectly converging the filters, which require pruning, to a single point. Finally, we demonstrate the superiority of the proposed method on various datasets and networks. In particular, on the ILSVRC-2012 dataset, our method removed 52.09% FLOPs with a negligible 0.15% top-1 accuracy drop on ResNet-50. As a result, we achieve the most outstanding performance compared to those reported in previous studies in terms of the trade-off between accuracy and computational complexity.

Sound-event recognition often utilizes time-frequency analysis to produce an image-like spectrogram that provides a rich visual representation of original signal in time and frequency. Convolutional Neural Networks (CNN) with the ability of learning discriminative spectrogram patterns are suitable for sound-event recognition. However, there is relatively little effort that CNN makes full use of the important temporal information. In this paper, we propose MCRNN, a Convolutional Recurrent Neural Networks (CRNN) architecture for sound-event recognition, the letter “M” in the name “MCRNN” of our model denotes the multi-sized convolution filters. Richer features are extracted by using several different convolution filter sizes at the last convolution layer. In addition, cochleagram images are used as the input layer of the network, instead of the traditional spectrogram image of a sound signal. Experiments on the RWCP dataset shows that the recognition rate of the proposed method achieved 98.4% in clean conditions, and it robustly outperforms the existing methods, the recognition rate increased by 0.9%, 1.9% and 10.3% in 20 dB, 10 dB and 0 dB signal-to-noise ratios (SNR), respectively.

Intra mode prediction for H.266/FVC video coding based on convolutional neural network

One of the ways to process an image presented in the form of a set of pixels, in order to further identify, classify the objects present on it is to display the specified set in the form of sets of certain features. Such features are not universal in nature, but rather significantly depend on the tasks under consideration. For certain classes of problems, such features (model) are selected that best allow the application of appropriate methods to solve the problem. The paper considers a class of problems for recognizing the emotional state on a person's face. In, a convolutional neural network (CNN) is used to detect emotions. CNN differs from multilayer perceptron (MLP) in that they have hidden layers called convolutional layers. The proposed method is based on a two-tier CNN system. At the first level, the background of the image is removed to better reflect emotions. A standard CNN network module is used to obtain the primary expression vector (EV). EV is formed by tracking the relevant important points of the face. EV is directly related to changes in facial expression. EV is obtained using a basic perceptron unit plotted on a face image with the background removed. In the proposed model at the last stage, there is a non-evolutionary perceptron layer. Each of the convoluted layers receives input data (images), converts them, and then takes them to the next level. After detecting a face, the CNN filter of the second part captures parts of the face, such as eyes, ears, lips, nose, and cheeks. The authors agree that the method has some limitations, and especially requires high computing power when setting up CNN. The technology of determination of characteristic features on the face for recognition of emotional manifestations is presented and experimentally investigated.

In the industry 4.0 era, video applications such as surveillance visual systems, video conferencing, or video broadcasting have been playing a vital role. In these applications, for manipulating and tracking objects in decoded video, the quality of decoded video should be consistent because it largely affects the performance of the machine analysis. To cope with this problem, we propose a novel perceptual video coding (PVC) solution in which a full reference quality metric named video multimethod assessment fusion (VMAF) is employed together with a deep convolutional neural network (CNN) to obtain consistent quality while still achieving high compression performance. First of all, in order to achieve the consistent quality requirement, we propose a CNN model with an expected VMAF as input to adaptively adjust the quantization parameters (QP) for each coding block. Afterwards, to increase the compression performance, a Lagrange coefficient of rate-distortion optimization (RDO) mechanism is adaptively computed according to rate-QP and quality-QP models. The experimental results show that the proposed PVC solution has achieved two targets simultaneously: the quality of video sequence is kept consistently with an expected quality level and the bit rate saving of the proposed method is higher than traditional video coding standards and the relevant benchmark, notably with around 10% bitrate saving on average.

Convolutional neural network (CNN) is the primary technique that has greatly promoted the development of computer vision technologies. However, there is little research on how to allocate parameters in different convolution layers when designing CNNs. We research mainly on revealing the relationship between CNN parameter distribution, i.e., the allocation of parameters in convolution layers, and the discriminative performance of CNN. Unlike previous works, we do not append more elements into the network, such as more convolution layers or denser short connections. We focus on enhancing the discriminative performance of CNN through varying its parameter distribution under strict size constraint. We propose an energy function to represent the CNN parameter distribution, which establishes the connection between the allocation of parameters and the discriminative performance of CNN. Extensive experiments with shallow CNNs on three public image classification data sets demonstrate that the CNN parameter distribution with a higher energy value will promote the model to obtain better performance. According to the motivated observation, the problem of finding the optimal parameter distribution can be transformed into an optimization problem of finding the biggest energy value. We present a simple yet effective guideline that uses balanced parameter distribution to design CNNs. Extensive experiments on ImageNet with three popular backbones, i.e., AlexNet, ResNet34, and ResNet101, demonstrate that the proposed guideline can make consistent improvements upon different baselines under strict size constraint.