Neural image compression leverages deep neural networks to outperform traditional image codecs in rate-distortion performance. However, the resulting models are also heavy, computationally demanding and generally optimized for a single rate, limiting their practical use. Focusing on practical image compression, we propose slimmable compressive autoencoders (SlimCAEs), where rate (R) and distortion (D) are jointly optimized for different capacities. Once trained, encoders and decoders can be executed at different capacities, leading to different rates and complexities. We show that a successful implementation of Slim-CAEs requires suitable capacity-specific RD tradeoffs. Our experiments show that SlimCAEs are highly flexible models that provide excellent rate-distortion performance, variable rate, and dynamic adjustment of memory, computational cost and latency, thus addressing the main requirements of practical image compression.

Class-incremental learning of deep networks sequentially increases the number of classes to be classified. During training, the network has only access to data of one task at a time, where each task contains several classes. In this setting, networks suffer from catastrophic forgetting which refers to the drastic drop in performance on previous tasks. The vast majority of methods have studied this scenario for classification networks, where for each new task the classification layer of the network must be augmented with additional weights to make room for the newly added classes. Embedding networks have the advantage that new classes can be naturally included into the network without adding new weights. Therefore, we study incremental learning for embedding networks. In addition, we propose a new method to estimate the drift, called semantic drift, of features and compensate for it without the need of any exemplars. We approximate the drift of previous tasks based on the drift that is experienced by current task data. We perform experiments on fine-grained datasets, CIFAR100 and ImageNet-Subset. We demonstrate that embedding networks suffer significantly less from catastrophic forgetting. We outperform existing methods which do not require exemplars and obtain competitive results compared to methods which store exemplars. Furthermore, we show that our proposed SDC when combined with existing methods to prevent forgetting consistently improves results.

Metric learning networks are used to compute image embeddings, which are widely used in many applications such as image retrieval and face recognition. In this paper, we propose to use network distillation to efficiently compute image embeddings with small networks. Network distillation has been successfully applied to improve image classification, but has hardly been explored for metric learning. To do so, we propose two new loss functions that model the communication of a deep teacher network to a small student network. We evaluate our system in several datasets, including CUB-200-2011, Cars-196, Stanford Online Products and show that embeddings computed using small student networks perform significantly better than those computed using standard networks of similar size. Results on a very compact network (MobileNet-0.25), which can be used on mobile devices, show that the proposed method can greatly improve Recall@1 results from 27.5\% to 44.6\%. Furthermore, we investigate various aspects of distillation for embeddings, including hint and attention layers, semi-supervised learning and cross quality distillation. (Code is available at https://github.com/yulu0724/EmbeddingDistillation).

The majority of existing color naming methods focuses on the eleven basic color terms of the English language. However, in many applications, different sets of color names are used for the accurate description of objects. Labeling data to learn these domain-specific color names is an expensive and laborious task. Therefore, in this article we aim to learn color names from weakly labeled data. For this purpose, we add an attention branch to the color naming network. The attention branch is used to modulate the pixel-wise color naming predictions of the network. In experiments, we illustrate that the attention branch correctly identifies the relevant regions. Furthermore, we show that our method obtains state-of-the-art results for pixel-wise and image-wise classification on the EBAY dataset and is able to learn color names for various domains.

Star images obtained by star sensors have a low signal‐to‐noise ratio due to various physical constraints. Low resolution also causes stellar centroid extraction error when traditional methods such as the Gaussian filter or adaptive median filter are utilised to de‐noise star images. An automatic centroid extraction method for noisy low‐resolution star images is proposed in this study. First, sparse representation is utilised to de‐noise the Poisson–Gaussian mixed noise of the low‐resolution star image. A high‐resolution star image is then reconstructed by using the low‐resolution sparse coefficient. Finally, the stellar centroids are extracted automatically by learning the relationship between the high‐resolution star image and corresponding stellar centroid image. Experimental results indicate that the positioning accuracy of the stellar centroids is also greatly enhanced by the reconstructed high‐resolution stellar centroid image. The correct rate of stellar centroid recognition is 99.35%; the positioning accuracy of stellar centroid and computing time are 16.21″ and 11.30 ms, respectively. The probability distributions of Poisson and Gaussian noises are 0.50 and 0.08, respectively, while the proposed method correctly recognises stellar centroids at a rate of 76.56%. The results presented here may provide a workable foundation for accurate attitude calculations of the celestial navigation system.

It is critical to classify the landing terrain from aerial images when an unmanned aerial vehicle lands at an unprepared site autonomously by using a vision sensor. Owing to the interference of illumination variations and noises, different terrains may show a similar image feature and the same terrain may have a different image feature, which brings great difficulties to image classification. To address this issue, a terrain classification method based on low-rank recovery and sparse representation is proposed. Color moments and Gabor texture feature are extracted and fused to construct a discriminative dictionary. Then, we perform low-rank matrix recovery for the dictionary by using augmented Lagrange multipliers and classify the test samples by sparse-representation-based classification. Experimental results on an aerial image database that we prepared by using the DJI Phantom 3 Advanced UAV verify the classification accuracy and robustness of the proposed method.

Protein-RNA interactions are vitally important to a number of fundamental cellular processes, including regulation of gene expression such as RNA splicing, transport and translation, protein synthesis and assembly of ribosome. More detailed information on the Protein-RNA interaction is helpful for comprehending the function notation and molecular regulatory mechanism, meanwhile, knowing the knowledge of Protein-RNA recognition can also help the biological scientist and researcher understand the site-directed mutagenesis and drug design. In the present work, we proposed a computational approach, based on SVM-KNN algorithm, with evolutionary information of spatial neighbour residues for prediction of protein-RNA interaction sites. The overall success rate obtained by 5-fold cross-validation is 78.00%, which is comparable or better than other existing methods, indicating our method is very promising for identifying and predicting protein-RNA interaction sites.

The estimation of Markov jump systems (MJS) is widely used in target tracking, fault detection, signal processing and digital communications. However, the above researches all assume that state measurement and additional mode observation are synchronous which means both state measurement and mode observation at each sampling time arrive at the fusion centre at the same time. The problem of estimation of MJS that mode observation is one-step lagged to its corresponding state measurement is considered. Along state-augmentation approach and the derivation of image-enhanced interacting multiple model (IE-IMM), a new generic estimation algorithm is proposed. It is shown by simulation result that the proposed algorithm is effective.

Multiple sensors image fusion is increasingly being employed in video surveillance. This paper presents a color based grayscale-fused image enhancement algorithm. The main advantage of this algorithm is to stand out the meaningful information of a certain sensor in the selected band of the color space, while maintaining the high resolution of the grayscale-fused image. Moreover, a performance evaluation method based on analyzing the average gradients of the color-fused image is presented and experiments are performed using images under complex environment. The results show that the addition of color to grayscale-fused image can significantly increase observer's sensitivity and the accuracy of further classification and recognition.