Richer Convolutional Features for Edge Detection Liu, Yun; Cheng, Ming-Ming; Hu, Xiaowei ...
IEEE transactions on pattern analysis and machine intelligence,
2019-Aug.-1, 2019-08-00, 2019-8-1, 20190801, Letnik:
41, Številka:
8
Journal Article
Recenzirano
Odprti dostop
Edge detection is a fundamental problem in computer vision. Recently, convolutional neural networks (CNNs) have pushed forward this field significantly. Existing methods which adopt specific layers ...of deep CNNs may fail to capture complex data structures caused by variations of scales and aspect ratios. In this paper, we propose an accurate edge detector using richer convolutional features (RCF). RCF encapsulates all convolutional features into more discriminative representation, which makes good usage of rich feature hierarchies, and is amenable to training via backpropagation. RCF fully exploits multiscale and multilevel information of objects to perform the image-to-image prediction holistically. Using VGG16 network, we achieve state-of-the-art performance on several available datasets. When evaluating on the well-known BSDS500 benchmark, we achieve ODS F-measure of 0.811 while retaining a fast speed (8 FPS). Besides, our fast version of RCF achieves ODS F-measure of 0.806 with 30 FPS. We also demonstrate the versatility of the proposed method by applying RCF edges for classical image segmentation.
Recent progress on saliency detection is substantial, benefiting mostly from the explosive development of Convolutional Neural Networks (CNNs). Semantic segmentation and saliency detection algorithms ...developed lately have been mostly based on Fully Convolutional Neural Networks (FCNs). There is still a large room for improvement over the generic FCN models that do not explicitly deal with the scale-space problem. Holisitcally-Nested Edge Detector (HED) provides a skip-layer structure with deep supervision for edge and boundary detection, but the performance gain of HED on saliency detection is not obvious. In this paper, we propose a new saliency method by introducing short connections to the skip-layer structures within the HED architecture. Our framework provides rich multi-scale feature maps at each layer, a property that is critically needed to perform segment detection. Our method produces state-of-the-art results on 5 widely tested salient object detection benchmarks, with advantages in terms of efficiency (0.08 seconds per image), effectiveness, and simplicity over the existing algorithms.
Rain is a common weather phenomenon, where object visibility varies with depth from the camera and objects faraway are visually blocked more by fog than by rain streaks. Existing methods and datasets ...for rain removal, however, ignore these physical properties, thereby limiting the rain removal efficiency on real photos. In this work, we first analyze the visual effects of rain subject to scene depth and formulate a rain imaging model collectively with rain streaks and fog; by then, we prepare a new dataset called RainCityscapes with rain streaks and fog on real outdoor photos. Furthermore, we design an end-to-end deep neural network, where we train it to learn depth-attentional features via a depth-guided attention mechanism, and regress a residual map to produce the rain-free image output. We performed various experiments to visually and quantitatively compare our method with several state-of-the-art methods to demonstrate its superiority over the others.
Shadow detection and shadow removal are fundamental and challenging tasks, requiring an understanding of the global image semantics. This paper presents a novel deep neural network design for shadow ...detection and removal by analyzing the spatial image context in a direction-aware manner. To achieve this, we first formulate the direction-aware attention mechanism in a spatial recurrent neural network (RNN) by introducing attention weights when aggregating spatial context features in the RNN. By learning these weights through training, we can recover direction-aware spatial context (DSC) for detecting and removing shadows. This design is developed into the DSC module and embedded in a convolutional neural network (CNN) to learn the DSC features at different levels. Moreover, we design a weighted cross entropy loss to make effective the training for shadow detection and further adopt the network for shadow removal by using a euclidean loss function and formulating a color transfer function to address the color and luminosity inconsistencies in the training pairs. We employed two shadow detection benchmark datasets and two shadow removal benchmark datasets, and performed various experiments to evaluate our method. Experimental results show that our method performs favorably against the state-of-the-art methods for both shadow detection and shadow removal.
In this paper, we propose an accurate edge detector using richer convolutional features (RCF). Since objects in natural images possess various scales and aspect ratios, learning the rich hierarchical ...representations is very critical for edge detection. CNNs have been proved to be effective for this task. In addition, the convolutional features in CNNs gradually become coarser with the increase of the receptive fields. According to these observations, we attempt to adopt richer convolutional features in such a challenging vision task. The proposed network fully exploits multiscale and multilevel information of objects to perform the image-to-image prediction by combining all the meaningful convolutional features in a holistic manner. Using VGG16 network, we achieve state-of-the-art performance on several available datasets. When evaluating on the well-known BSDS500 benchmark, we achieve ODS F-measure of 0.811 while retaining a fast speed (8 FPS). Besides, our fast version of RCF achieves ODS F-measure of 0.806 with 30 FPS.
Putting the N in nanotube: Carbon nanotubes (CNTs) catalyze the aerobic oxidation of cyclohexane into cyclohexanol, cyclohexanone, and adipic acid with excellent activity and controllable ...selectivity. The catalytic activity is further enhanced by nitrogen dopants in the nanotube (see diagram; AC=activated carbon, MWCNT=multiwalled CNT, N‐CNT=nitrogen‐doped CNT).
Inverse synthetic aperture radar (ISAR) imaging of a precessing target, which is a kind of fast spinning target, is faced with migration through range cell when using traditional imaging algorithms. ...Theory of compressed sensing (CS) suggests that exact recovery of an unknown sparse signal with an overwhelming probability can be achieved from very limited number of samples. A cycle shift smoothed L0 algorithm based on CS is proposed in this paper for high-resolution ISAR imaging of precessing targets by exploiting sparse apertures. A precessing cone-shaped target model is built and a 3-D reconstruction method based on multistatic ISAR is proposed. Simulations and electromagnetic computation verify the validity of the proposed method.
Extending cycling distances is crucial for sustainable urban transport development and plays a role in encouraging the shift from motorized vehicles to public transport. However, there is a lack of ...research examining the combined impacts of both objective and perceived aspects of the cycling environment on cycling distance, and the existence of threshold effects remains unclear. This study uses 2019 cycling data from Shenzhen, China, employing the XGBoost algorithm to uncover the relative importance and thresholds of objective and perceived factors in the cycling environment. The results indicate that population density (24.8%), road network density (15.2%), the proportion of recreational facilities (9.1%), perceived accessibility (8.0%), and comfort (8.6%) hold high relative importance in predicting cycling distance. Also, maintaining road network density between 3 to 6 km/km2 and increasing the population density to exceed 22,000 people/km2 proves effective in extending cycling distances. Land use demonstrates a threshold effect, with cycling distances increasing when the recreational facilities share exceeds 8%, transport facilities share remains below 25%, and commercial facilities share stays below 30%. Perceived metrics exhibit a clear threshold effect. The study identifies that perceived safety indicates a psychological bottleneck in increasing cycling distance. Perceived accessibility is positively correlated with cycling distance when accessibility is at a low level, while comfort shows a positive correlation with cycling distance when comfort is at a high level. These findings can contribute to refining land planning and prioritizing resource allocation for organizations aiming to promote non-motorized travel and design bicycle-friendly environments.
In order to explore the changes that autonomous vehicles on the road would bring to the current traffic and make full use of the intelligent features of autonomous vehicles, the article defines a ...self-balancing system of autonomous vehicles. Based on queuing theory and stochastic process, the self-balancing system model with self-balancing characteristics is established to balance the utilization rate of autonomous vehicles under the conditions of ensuring demand and avoiding an uneven distribution of vehicle resources in the road network. The performance indicators of the system are calculated by the MVA (Mean Value Analysis) method. The analysis results show that the self-balancing process could reduce the average waiting time of customers significantly in the system, alleviate the service pressure while ensuring travel demand, fundamentally solve the phenomenon of concentrated idleness after the use of vehicles in the current traffic, maximize the use of the mobile vehicles in the system, and realize the self-balancing of the traffic network while reducing environmental pollution and saving energy.