No abstract

Pedestrian analysis plays a vital role in intelligent video surveillance and is a key component for security-centric computer vision systems. Despite that the convolutional neural networks are remarkable in learning discriminative features from images, the learning of comprehensive features of pedestrians for fine-grained tasks remains an open problem. In this study, we propose a new attentionbased deep neural network, named as HydraPlus-Net (HPnet), that multi-directionally feeds the multi-level attention maps to different feature layers. The attentive deep features learned from the proposed HP-net bring unique advantages: (1) the model is capable of capturing multiple attentions from low-level to semantic-level, and (2) it explores the multi-scale selectiveness of attentive features to enrich the final feature representations for a pedestrian image. We demonstrate the effectiveness and generality of the proposed HP-net for pedestrian analysis on two tasks, i.e. pedestrian attribute recognition and person reidentification. Intensive experimental results have been provided to prove that the HP-net outperforms the state-of-theart methods on various datasets.

No abstract

Assessing potential future changes in arctic and boreal plant species productivity, ecosystem composition, and canopy complexity is essential for understanding environmental responses under expected altered climate forcing. We examined potential changes in the dominant plant functional types (PFTs) of the sedge tundra, shrub tundra, and boreal forest ecosystems in ecotonal northern Alaska, USA, for the years 2003-2100. We compared energy feedbacks associated with increases in biomass to energy feedbacks associated with changes in the duration of the snow-free season. We based our simulations on nine input climate scenarios from the Intergovernmental Panel on Climate Change (IPCC) and a new version of the Terrestrial Ecosystem Model (TEM) that incorporates biogeochemistry, vegetation dynamics for multiple PFTs (e.g., trees, shrubs, grasses, sedges, mosses), multiple vegetation pools, and soil thermal regimes. We found mean increases in net primary productivity (NPP) in all PFTs. Most notably, birch (Betula spp.) in the shrub tundra showed increases that were at least three times larger than any other PFT. Increases in NPP were positively related to increases in growing-season length in the sedge tundra, but PFTs in boreal forest and shrub tundra showed a significant response to changes in light availability as well as growing-season length. Significant NPP responses to changes in vegetation uptake of nitrogen by PFT indicated that some PFTs were better competitors for nitrogen than other PFTs. While NPP increased, heterotrophic respiration (RH) also increased, resulting in decreases or no change in net ecosystem carbon uptake. Greater aboveground biomass from increased NPP produced a decrease in summer albedo, greater regional heat absorption (0.34 +/- 0.23 W x m(-2) x 10 yr(-1) [mean +/- SD]), and a positive feedback to climate warming. However, the decrease in albedo due to a shorter snow season (-5.1 +/- 1.6 d/10 yr) resulted in much greater regional heat absorption (3.3 +/- 1.24 W x m(-2) x 10 yr(-1)) than that associated with increases in vegetation. Through quantifying feedbacks associated with changes in vegetation and those associated with changes in the snow season length, we can reach a more integrated understanding of the manner in which climate change may impact interactions between high-latitude ecosystems and the climate system.

Vehicle re-identification is an important problem and has many applications in video surveillance and intelligent transportation. It gains increasing attention because of the recent advances of person re-identification techniques. However, unlike person re-identification, the visual differences between pairs of vehicle images are usually subtle and even challenging for humans to distinguish. Incorporating additional spatio-temporal information is vital for solving the challenging re-identification task. Existing vehicle re-identification methods ignored or used oversimplified models for the spatio-temporal relations between vehicle images. In this paper, we propose a two-stage framework that incorporates complex spatio-temporal information for effectively regularizing the re-identification results. Given a pair of vehicle images with their spatiotemporal information, a candidate visual-spatio-temporal path is first generated by a chain MRF model with a deeply learned potential function, where each visual-spatiotemporal state corresponds to an actual vehicle image with its spatio-temporal information. A Siamese-CNN+Path-LSTM model takes the candidate path as well as the pairwise queries to generate their similarity score. Extensive experiments and analysis show the effectiveness of our proposed method and individual components.