Person re-identification (re-ID) is a fundamental problem in the field of computer vision. The pe... more Person re-identification (re-ID) is a fundamental problem in the field of computer vision. The performance of deep learning-based person re-ID models suffers from a lack of training data. In this work, we introduce a novel image-specific data augmentation method on the feature map level to enforce feature diversity in the network. Furthermore, an attention assignment mechanism is proposed to enforce that the person re-ID classifier focuses on nearly all important regions of the input person image. To achieve this, a three-stage framework is proposed. First, a baseline classification network is trained for person re-ID. Second, an attention assignment network is proposed based on the baseline network, in which the attention module learns to suppress the response of the current detected regions and re-assign attentions to other important locations. By this means, multiple important regions for classification are highlighted by the attention map. Finally, the attention map is integrate...
IEEE transactions on image processing : a publication of the IEEE Signal Processing Society, Jan 6, 2017
Benefiting from global rank constraints, the lowrank representation (LRR) method has been shown t... more Benefiting from global rank constraints, the lowrank representation (LRR) method has been shown to be an effective solution to subspace learning. However, the global mechanism also means that the LRR model is not suitable for handling large-scale data or dynamic data. For large-scale data, the LRR method suffers from high time complexity, and for dynamic data, it has to recompute a complex rank minimization for the entire data set whenever new samples are dynamically added, making it prohibitively expensive. Existing attempts to online LRR either take a stochastic approach or build the representation purely based on a small sample set and treat new input as out-of-sample data. The former often requires multiple runs for good performance and thus takes longer time to run, and the latter formulates online LRR as an out-ofsample classification problem and is less robust to noise. In this paper, a novel online low-rank representation subspace learning method is proposed for both large-s...
2014 5th International Conference on Digital Home, 2014
Saliency detection is an important task to detect humans' visual attention region, plenty of ... more Saliency detection is an important task to detect humans' visual attention region, plenty of methods are proposed to solve it. Most of them use single level pixels or patches to reveal the structure of the given image. However, the complexity of background regions and the unevenness of foreground objects still challenge these methods. To distinguish the saliency object from the complex background region and guarantee the uniformity of patches in the same object. A novel saliency detection method based sub modular optimization is proposed in this paper. Firstly we propose an image structure with both mid-level super pixels and object-level regions. The object-level regions are created by optimal sub modular clustering. Secondly, we use spatial based color contrast method to calculate the raw saliency map. Then the raw saliency is adjusted by three priors. The priors can reduce the impact of complex background regions and help generating the mid-level saliency map. Finally, the sub modular optimization method is used to jointly select the diffusion seeds and diffuse the mid-level saliency map into object-level saliency map.
2013 IEEE International Conference on Image Processing, 2013
ABSTRACT Conventional saliency detection approaches are human fixation detection and single domin... more ABSTRACT Conventional saliency detection approaches are human fixation detection and single dominant region detection. However, real-world photographs usually consist of multiple dominant regions. We propose a saliency detection method with the aim to highlight objects as a whole and distinguish objects with different saliency levels. It combines the bottom-up approach and top-down approach via two nested levels of hierarchical segmentations - the coarse level objects and fine level details. We first calculate a preliminary saliency on the fine patches with a random walk model. Then a location cue and an object-level cue are fused to refine the preliminary saliency to emphasize the objects against the background. At last, the object-level saliency map is synthesized via a heat diffusion process restricted by the coarse level patches to enhance object saliency and distinguish saliency between different objects. Extensive evaluation on a publicly available database verifies that our method outperforms the state-of-the-art algorithms.
Person re-identification (re-ID) is a fundamental problem in the field of computer vision. The pe... more Person re-identification (re-ID) is a fundamental problem in the field of computer vision. The performance of deep learning-based person re-ID models suffers from a lack of training data. In this work, we introduce a novel image-specific data augmentation method on the feature map level to enforce feature diversity in the network. Furthermore, an attention assignment mechanism is proposed to enforce that the person re-ID classifier focuses on nearly all important regions of the input person image. To achieve this, a three-stage framework is proposed. First, a baseline classification network is trained for person re-ID. Second, an attention assignment network is proposed based on the baseline network, in which the attention module learns to suppress the response of the current detected regions and re-assign attentions to other important locations. By this means, multiple important regions for classification are highlighted by the attention map. Finally, the attention map is integrate...
IEEE transactions on image processing : a publication of the IEEE Signal Processing Society, Jan 6, 2017
Benefiting from global rank constraints, the lowrank representation (LRR) method has been shown t... more Benefiting from global rank constraints, the lowrank representation (LRR) method has been shown to be an effective solution to subspace learning. However, the global mechanism also means that the LRR model is not suitable for handling large-scale data or dynamic data. For large-scale data, the LRR method suffers from high time complexity, and for dynamic data, it has to recompute a complex rank minimization for the entire data set whenever new samples are dynamically added, making it prohibitively expensive. Existing attempts to online LRR either take a stochastic approach or build the representation purely based on a small sample set and treat new input as out-of-sample data. The former often requires multiple runs for good performance and thus takes longer time to run, and the latter formulates online LRR as an out-ofsample classification problem and is less robust to noise. In this paper, a novel online low-rank representation subspace learning method is proposed for both large-s...
2014 5th International Conference on Digital Home, 2014
Saliency detection is an important task to detect humans' visual attention region, plenty of ... more Saliency detection is an important task to detect humans' visual attention region, plenty of methods are proposed to solve it. Most of them use single level pixels or patches to reveal the structure of the given image. However, the complexity of background regions and the unevenness of foreground objects still challenge these methods. To distinguish the saliency object from the complex background region and guarantee the uniformity of patches in the same object. A novel saliency detection method based sub modular optimization is proposed in this paper. Firstly we propose an image structure with both mid-level super pixels and object-level regions. The object-level regions are created by optimal sub modular clustering. Secondly, we use spatial based color contrast method to calculate the raw saliency map. Then the raw saliency is adjusted by three priors. The priors can reduce the impact of complex background regions and help generating the mid-level saliency map. Finally, the sub modular optimization method is used to jointly select the diffusion seeds and diffuse the mid-level saliency map into object-level saliency map.
2013 IEEE International Conference on Image Processing, 2013
ABSTRACT Conventional saliency detection approaches are human fixation detection and single domin... more ABSTRACT Conventional saliency detection approaches are human fixation detection and single dominant region detection. However, real-world photographs usually consist of multiple dominant regions. We propose a saliency detection method with the aim to highlight objects as a whole and distinguish objects with different saliency levels. It combines the bottom-up approach and top-down approach via two nested levels of hierarchical segmentations - the coarse level objects and fine level details. We first calculate a preliminary saliency on the fine patches with a random walk model. Then a location cue and an object-level cue are fused to refine the preliminary saliency to emphasize the objects against the background. At last, the object-level saliency map is synthesized via a heat diffusion process restricted by the coarse level patches to enhance object saliency and distinguish saliency between different objects. Extensive evaluation on a publicly available database verifies that our method outperforms the state-of-the-art algorithms.
Uploads