Search

A biologically inspired spatiotemporal saliency attention model based on entropy value is proposed in this paper. This model includes a dynamic attention phase and a static attention phase. In the dynamic attention phase, low-level visual features are extracted from current and some previous frames. Every feature map is resized into some different sizes. The feature maps in same size and same feature for all the frames are used to calculate the entropy value map. All the entropy maps are normalized and are fused into a dynamic saliency map. In the static attention phase, same features are extracted and form multi-scale feature maps by center-surround differences in current frame, and then those feature maps are transformed into conspicuity maps, which are linearly combined into a static saliency map. Our model decides salient regions based on a spatiotemporal saliency map which is generated by integration of the dynamic and the static saliency map. Experimental results indicate that: when there is noise among the frames or there is change of illumination among the frames, our model is excellent to Shi's model and Marat's model; when the moving objects do not belong to the static salient regions, our model is better than Ban's model. [ABSTRACT FROM AUTHOR]

• Analysing spatial effects of trees is critical for improving forest management. • This study explored spatial effects of trees using biomass data in Yunnan, China. • The significant spatial autocorrelations vary with stands, components, and distances. • The stands' spatial heterogeneity increased with distance and stabilised at 10 m. The trees in a given stand compete with each other for light, water and nutrients, producing spatial effects. To understand spatial effects, it is critical to characterise the spatial distributions and patterns of the aboveground biomass (AGB) of individual trees and the biomass of their components (stem, bark, branches and foliage). This study investigated the spatial effects by examining the AGB of individual trees and the biomass of their components in the context of typical sub-tropical tree species in Yunnan, China, including a Pinus kesiya var. langbianensis natural forest (PN), a Pinus kesiya var. langbianensis plantation (PP), and a Eucalyptus spp. plantation (EP), using datasets for three clear-cutting plots. Ripley's L function was employed to identify the spatial distributions and patterns of the AGB of individual trees and the biomass of their components. Then, Global and Local Moran's I indices were utilised to analyse spatial autocorrelation, with Intra-group variance being calculated to quantify spatial heterogeneity. This led to following findings. Firstly, with increasing distance, the spatial heterogeneity of the three stands increased before stabilising at a distance of 10 m. Secondly, excluding bark, all components of the tree biomass in the PN stand exhibited significant dispersion. In contrast, all the components in the PP stand showed significant dispersion. The spatial distributions and patterns of trees and their biomass in the EP stand indicated the coexistence of clustering and dispersion. Moreover, the significant spatial autocorrelations of tree AGB and components biomass could be noticed but depending on the stands, components and distances. These findings provide theoretical support for sustainable forest management based on biomass and carbon stocks. [ABSTRACT FROM AUTHOR]

Abstract: Automatic image segmentation is always a fundamental but challenging problem in computer vision. The simplest approach to image segmentation may be clustering feature vectors of pixels at first, then labeling each pixel with its corresponding cluster. This requires that the clustering on feature space must be robust. However, most of popular clustering algorithms could not obtain a robust clustering result yet, if the clusters in feature space have a complex distribution. Generally, for most of clustering-based segmentation methods, it still needs more constraints of positional relations between pixels in image lattice to be utilized during the procedure of clustering. Our works in this paper address the problem of image segmentation under the paradigm of pure clustering-then-labeling. A robust clustering algorithm which could maintain good coherence of data in feature space is proposed and utilized to do clustering on the L∗a∗b∗ color feature space of pixels. Image segmentation is straightforwardly obtained by setting each pixel with its corresponding cluster. Further, based on the theory of Minimum Description Length, an effective approach to automatic parameter selection for our segmentation method is also proposed. We test our segmentation method on Berkeley segmentation database, and the experimental results show that our method compares favorably against some state-of-the-art segmentation methods. [Copyright &y& Elsevier]

• We propose a domain-adaptive video object detection framework to generate scene-specific pedestrian detectors in different scenarios without human-labeled target domain samples. • We propose a cycle semantic transfer network to achieve adversarial domain adaptation across different surveillance videos. • We propose an online gradual optimization algorithm to iteratively specialize a generic detector to a target domain. This algorithm exhibits better fault tolerance for mislabeled samples by simulating an active learning process. • We conduct experiments on multiple datasets to evaluate the performance of the algorithm. The proposed self-learning method performs even better than supervised learning methods and existing scene-specific pedestrian detection methods. Objects from one category may be drawn from different distributions due to diverse illuminations, backgrounds, and camera viewpoints. Traditional object detection methods generally perform poorly due to the domain shift. To address this problem, we propose to train a domain-adaptive scene-specific pedestrian detector in an unsupervised manner. A generic detector is transferred to different target domains from one labeled source domain dataset without human-annotated target samples. Specifically, we first extend the generic detector to a dual-boundary classifier and collect hard samples as unlabeled target samples according to the detection confidence. Then, we propose a cycle semantic transfer network to align the instance-level and class-level distributions between the source domain and target domain and automatically label the hard samples. The initial generic detector is then re-trained by these labeled hard samples and specialized to a target scene. This process can be conveniently extended to different surveillance scenarios and generate specific detectors under various static camera viewpoints. Moreover, to reduce the impact of mislabeled hard samples on the generic detector, an online gradual optimization algorithm is proposed to iteratively update the generic model, thereby obtaining an optimized process that is insensitive to individual mislabeled target samples. Extensive experiments show that even if the target domain is not manually annotated, the proposed self-learning method demonstrates the effectiveness of pedestrian detection in various domain shift scenarios, and it outperforms existing scene-specific pedestrian detection methods and some classic supervised methods. [ABSTRACT FROM AUTHOR]