Showing papers by "Ning Zhang published in 2015"
07 Jun 2015
TL;DR: In this paper, a pose-invariant PErson Recognition (PIPER) method is proposed, which accumulates the cues of poselet-level person recognizers trained by deep convolutional networks to discount for the pose variations, combined with a face recognizer and a global recognizer.
Abstract: We explore the task of recognizing peoples' identities in photo albums in an unconstrained setting. To facilitate this, we introduce the new People In Photo Albums (PIPA) dataset, consisting of over 60000 instances of ∼2000 individuals collected from public Flickr photo albums. With only about half of the person images containing a frontal face, the recognition task is very challenging due to the large variations in pose, clothing, camera viewpoint, image resolution and illumination. We propose the Pose Invariant PErson Recognition (PIPER) method, which accumulates the cues of poselet-level person recognizers trained by deep convolutional networks to discount for the pose variations, combined with a face recognizer and a global recognizer. Experiments on three different settings confirm that in our unconstrained setup PIPER significantly improves on the performance of DeepFace, which is one of the best face recognizers as measured on the LFW dataset.
132 citations
Posted Content•
TL;DR: The Pose Invariant PErson Recognition (PIPER) method is proposed, which accumulates the cues of poselet-level person recognizers trained by deep convolutional networks to discount for the pose variations, combined with a face recognizer and a global recognizer.
Abstract: We explore the task of recognizing peoples' identities in photo albums in an unconstrained setting. To facilitate this, we introduce the new People In Photo Albums (PIPA) dataset, consisting of over 60000 instances of 2000 individuals collected from public Flickr photo albums. With only about half of the person images containing a frontal face, the recognition task is very challenging due to the large variations in pose, clothing, camera viewpoint, image resolution and illumination. We propose the Pose Invariant PErson Recognition (PIPER) method, which accumulates the cues of poselet-level person recognizers trained by deep convolutional networks to discount for the pose variations, combined with a face recognizer and a global recognizer. Experiments on three different settings confirm that in our unconstrained setup PIPER significantly improves on the performance of DeepFace, which is one of the best face recognizers as measured on the LFW dataset.
117 citations
Posted Content•
TL;DR: This work unifies steps in an end-to-end trainable network supervised by keypoint locations and class labels that localizes parts by a fully convolutional network to focus the learning of feature representations for the fine-grained classification task.
Abstract: Pose variation and subtle differences in appearance are key challenges to fine-grained classification. While deep networks have markedly improved general recognition, many approaches to fine-grained recognition rely on anchoring networks to parts for better accuracy. Identifying parts to find correspondence discounts pose variation so that features can be tuned to appearance. To this end previous methods have examined how to find parts and extract pose-normalized features. These methods have generally separated fine-grained recognition into stages which first localize parts using hand-engineered and coarsely-localized proposal features, and then separately learn deep descriptors centered on inferred part positions. We unify these steps in an end-to-end trainable network supervised by keypoint locations and class labels that localizes parts by a fully convolutional network to focus the learning of feature representations for the fine-grained classification task. Experiments on the popular CUB200 dataset show that our method is state-of-the-art and suggest a continuing role for strong supervision.
72 citations
Posted Content•
TL;DR: This paper proposed two compact bilinear representations with the same discriminative power as the full Bilinear representation but with only a few thousand dimensions, which allow back-propagation of classification errors enabling an end-to-end optimization of the visual recognition system.
Abstract: Bilinear models has been shown to achieve impressive performance on a wide range of visual tasks, such as semantic segmentation, fine grained recognition and face recognition. However, bilinear features are high dimensional, typically on the order of hundreds of thousands to a few million, which makes them impractical for subsequent analysis. We propose two compact bilinear representations with the same discriminative power as the full bilinear representation but with only a few thousand dimensions. Our compact representations allow back-propagation of classification errors enabling an end-to-end optimization of the visual recognition system. The compact bilinear representations are derived through a novel kernelized analysis of bilinear pooling which provide insights into the discriminative power of bilinear pooling, and a platform for further research in compact pooling methods. Experimentation illustrate the utility of the proposed representations for image classification and few-shot learning across several datasets.
34 citations
01 Jan 2015
TL;DR: This work proposes pose-normalized representations, which align training exemplars, either piecewise by part or globally for the whole object, effectively factoring out differences in pose and in camera viewing angle, and introduces the part-based RCNN method as an extension of state-of-art object detection method RCNN for fine-grained categorization.
Abstract: In contrast to basic-level object recognition, fine-grained categorization aims to distinguishbetween subordinate categories, such as different animal breeds or species, plant species or man-made product models. The problem can be extremely challenging due to the subtle differences in the appearance of certain parts across related categories and often requires distinctions that must be conditioned on the object pose for reliable identification. Discriminative markings are often highly localized, leading traditional object recognition approaches to struggle with the large pose variations often present in these domains. Face recognition is the classic case of fine-grained recognition, and it is noteworthy that the best face recognition methods jointly discover facial landmarks and extract features from those locations. We propose pose-normalized representations, which align training exemplars, either piecewise by part or globally for the whole object, effectively factoring out differences in pose and in camera viewing angle.I first present the methods of using the idea of pose-normalization for two related applications: human attribute classification and person recognition beyond frontal face. Following the recent success of deep learning, we use deep convolutional features as feature representations. Next, I will introduce the part-based RCNN method as an extension of state-of-art object detection method RCNN for fine-grained categorization. The model learns both whole-object and part detectors, and enforces learned geometric constraints between them. I will also show the results of using the recent compact bilinear features to generate the pose-normalized representations. However, bottom-up region proposals is limited by handengineered features and in the final work, I will present a fully convolution deep network, trained end-to-end for part localization and fine-grained classification.
1 citations