•The first comprehensive survey on deep-learning-based trackers.•Review existing deep visual trackers from three different perspectives.•Large-scale benchmark evaluations of deep visual ...trackers.•Summarize cutting-edge research works and discuss future directions•Provide useful insights and conclusions for deep visual trackers.
Recently, deep learning has achieved great success in visual tracking. The goal of this paper is to review the state-of-the-art tracking methods based on deep learning. First, we introduce the background of deep visual tracking, including the fundamental concepts of visual tracking and related deep learning algorithms. Second, we categorize the existing deep-learning-based trackers into three classes according to network structure, network function and network training. For each categorize, we explain its analysis of the network perspective and analyze papers in different categories. Then, we conduct extensive experiments to compare the representative methods on the popular OTB-100, TC-128 and VOT2015 benchmarks. Based on our observations, we conclude that: (1) The usage of the convolutional neural network (CNN) model could significantly improve the tracking performance. (2) The trackers using the convolutional neural network (CNN) model to distinguish the tracked object from its surrounding background could get more accurate results, while using the CNN model for template matching is usually faster. (3) The trackers with deep features perform much better than those with low-level hand-crafted features. (4) Deep features from different convolutional layers have different characteristics and the effective combination of them usually results in a more robust tracker. (5) The deep visual trackers using end-to-end networks usually perform better than the trackers merely using feature extraction networks. (6) For visual tracking, the most suitable network training method is to per-train networks with video information and online fine-tune them with subsequent observations. Finally, we summarize our manuscript and highlight our insights, and point out the further trends for deep visual tracking.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UL, UM, UPCLJ, UPUK, ZRSKP
Automated optical inspection of FAST is realized by exploiting advances in drone technology and deep-learning techniques. The AI-powered drone-based automated inspection is time-efficient and ...reliable, which guarantees the stable operation of FAST.
This work investigates the enhancement of dielectric properties of La2NiO4-CuO terpolymer nanocomposites using surface modified SiC nanoparticles as fillers. SiC nanoparticles were subjected to acid ...treatment and annealing at 1000 °C to tailor the surface chemistry. Acid treatment involved immersing the SiC powder in a mixture of H2SO4 and HNO3 acids and sonicating for 30 min, followed by stirring at 60 °C for 6 h. Annealing was carried out at 1000 °C for 2 h in an inert argon atmosphere. La2NiO4 and CuO nanoparticles were synthesized by sol-gel and precipitation techniques respectively, with average particle sizes of 20–40 nm and 10–15 nm confirmed by XRD and FESEM. Nanocomposites were fabricated by dispersing 2–10 wt% of untreated, acid treated and annealed SiC nanoparticles in a La2NiO4-CuO mixture solution, using PVDF as the polymer matrix. Impedance spectroscopy revealed that addition of 5 wt% acid treated SiC resulted in the highest dielectric constant of 42 at 1 kHz, in comparison to 25 for the unfilled polymer. This was attributed to increased interfacial polarization arising from uniform dispersion of nanoparticles and abundant charge trapping sites introduced by the SiC filler.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
Neonatal hypoxic ischemic encephalopathy is a common disease, which is caused by fetal hypoxia, asphyxia, and other reasons. It may cause sequelae of the nervous system and even death in children. ...Computer tomography examination can clarify the scope and location of the disease and provide the basis for clinical treatment and prognosis. Relevant personnel analyzed the symptoms of ischemic hypoxia and found that ischemia and hypoxia were the main causes of encephalopathy. Neonatal ischemia and hypoxia are easy to cause serious damage. At present, with the development of medicine, the function of the human brain is the most important issue in natural science. The law of neural activity and the role of molecular cells, organs, and systems have fundamental construction significance for the prevention and treatment of nerve and mental diseases. By analyzing the value of the diagnosis of neonatal hypoxic-ischemic encephalopathy in the analysis of experimental data, by setting the newborns in the controlled group and the trial group as experimental subjects, this paper analyzed the situation of newborns in terms of body temperature, body weight, and respiratory rate, and used Apgar score to score these standards. It was found that the score of the controlled group was 7 and above, and the score of the trial group was below 7. It was found that the Apgar scoring method was more simple. Then, the newborns were analyzed by cord blood gas analysis. It was found that most of the data in the control group were between 7.8 and 8.4, and the data in the trial group were between 5.8 and 7.1. The umbilical blood gas analysis score of the experimental group was lower than that of the control group. By comparing the satisfaction of cord blood gas analysis and the Apgar score, it was found that the satisfaction of cord blood gas analysis was 24.06% higher than that of the Apgar score.
C‐dots on hand: Luminescent carbon nanodots were synthesized and were shown to be biocompatible, have low toxicity, and distinctive photoluminescence properties. These C‐dots are inexpensive to ...synthesize and could potentially be used for versatile applications, such as anticounterfeiting, information encryption, and information storage.
Full text
Available for:
BFBNIB, FZAB, GIS, IJS, KILJ, NLZOH, NUK, OILJ, SAZU, SBCE, SBMB, UL, UM, UPUK
Visual Tracking with Fully Convolutional Networks Wang, Lijun; Ouyang, Wanli; Wang, Xiaogang ...
2015 IEEE International Conference on Computer Vision (ICCV),
12/2015
Conference Proceeding, Journal Article
We propose a new approach for general object tracking with fully convolutional neural network. Instead of treating convolutional neural network (CNN) as a black-box feature extractor, we conduct ...in-depth study on the properties of CNN features offline pre-trained on massive image data and classification task on ImageNet. The discoveries motivate the design of our tracking system. It is found that convolutional layers in different levels characterize the target from different perspectives. A top layer encodes more semantic features and serves as a category detector, while a lower layer carries more discriminative information and can better separate the target from distracters with similar appearance. Both layers are jointly used with a switch mechanism during tracking. It is also found that for a tracking target, only a subset of neurons are relevant. A feature map selection method is developed to remove noisy and irrelevant feature maps, which can reduce computation redundancy and improve tracking accuracy. Extensive evaluation on the widely used tracking benchmark 36 shows that the proposed tacker outperforms the state-of-the-art significantly.
Silicon (Si) alleviates cadmium (Cd) toxicity in rice (Oryza sativa). However, the chemical mechanisms at the single‐cell level are poorly understood. Here, a suspension of rice cells exposed to Cd ...and/or Si treatments was investigated using a combination of plant cell nutritional, molecular biological, and physical techniques including in situ noninvasive microtest technology (NMT), polymerase chain reaction (PCR), inductively coupled plasma mass spectroscopy (ICP‐MS), and atomic force microscopy (AFM) in Kelvin probe mode (KPFM). We found that Si‐accumulating cells had a significantly reduced net Cd²⁺influx, compared with that in Si‐limited cells. PCR analyses of the expression levels of Cd and Si transporters in rice cells showed that, when the Si concentration in the medium was increased, expression of the Si transporter gene Low silicon rice 1 (Lsi1) was up‐regulated, whereas expression of the gene encoding the transporter involved in the transport of Cd, Natural resistance‐associated macrophage protein 5 (Nramp5), was down‐regulated. ICP‐MS results revealed that 64% of the total Si in the cell walls was bound to hemicellulose constituents following the fractionation of the cell walls, and consequently inhibited Cd uptake. Furthermore, AFM in KPFM demonstrated that the heterogeneity of the wall surface potential was higher in cells cultured in the presence of Si than in those cultured in its absence, and was homogenized after the addition of Cd. These results suggest that a hemicellulose‐bound form of Si with net negative charges is responsible for inhibition of Cd uptake in rice cells by a mechanism of Si‐hemicellulose matrixCd complexation and subsequent co‐deposition.
Full text
Available for:
BFBNIB, FZAB, GIS, IJS, KILJ, NLZOH, NMLJ, NUK, OILJ, PNG, SAZU, SBCE, SBMB, UL, UM, UPUK
Deep Neural Networks (DNNs) have substantially improved the state-of-the-art in salient object detection. However, training DNNs requires costly pixel-level annotations. In this paper, we leverage ...the observation that image-level tags provide important cues of foreground salient objects, and develop a weakly supervised learning method for saliency detection using image-level tags only. The Foreground Inference Network (FIN) is introduced for this challenging task. In the first stage of our training method, FIN is jointly trained with a fully convolutional network (FCN) for image-level tag prediction. A global smooth pooling layer is proposed, enabling FCN to assign object category tags to corresponding object regions, while FIN is capable of capturing all potential foreground regions with the predicted saliency maps. In the second stage, FIN is fine-tuned with its predicted saliency maps as ground truth. For refinement of ground truth, an iterative Conditional Random Field is developed to enforce spatial label consistency and further boost performance. Our method alleviates annotation efforts and allows the usage of existing large scale training sets with image-level tags. Our model runs at 60 FPS, outperforms unsupervised ones with a large margin, and achieves comparable or even superior performance than fully supervised counterparts.