This paper considers learning deep features from longtailed data. We observe that in the deep feature space, the head classes and the tail classes present different distribution patterns. The head classes have a relatively large spatial span, while the tail classes have significantly small spatial span, due to the lack of intra-class diversity. This uneven distribution between head and tail classes distorts the overall feature space, which compromises the discriminative ability of the learned features. Intuitively, we seek to expand the distribution of the tail classes by transferring from the head classes, so as to alleviate the distortion of the feature space. To this end, we propose to construct each feature into a " feature cloud". If a sample belongs to a tail class, the corresponding feature cloud will have relatively large distribution range, in compensation to its lack of diversity. It allows each tail sample to push the samples from other classes far away, recovering the intra-class diversity of tail classes. Extensive experimental evaluations on person re-identification and face recognition tasks confirm the effectiveness of our method.

No abstract

In person re-IDentification (re-ID) task, the learning of part-level features benefits from fine-grained information. To facilitate part alignment, which is a prerequisite for learning part-level features, a popular approach is to detect semantic parts with the use of human parsing or pose estimation. Such methods of semantic partition do offer cues to good part alignment but are prone to noisy part detection, especially when they are employed in an off-the-shelf manner. In response, this paper proposes a novel part feature learning method for re-ID, that suppresses the impact of noisy semantic part detection through Supervised Non-local Similarity (SNS) learning. Given several detected semantic parts, SNS first locates their center points on the convolutional feature maps for use as a set of anchors and then evaluates the similarity values between these anchors and each pixel on the feature maps. The non-local similarity learning is supervised such that: each anchor should be similar to itself and simultaneously dissimilar to any other anchors, thus yielding the SNS. Finally, each anchor absorbs features from all of the similar pixels on the convolutional feature maps to generate a corresponding part feature (SNS feature). We evaluate our method with extensive experiments conducted under both holistic and partial re-ID scenarios. Experimental results confirm that SNS consistently improves re-ID accuracy using human parsing or pose estimation, and that our results are on par with state-of-the-art methods.

No abstract

Current person image retrieval methods have achieved great improvements in accuracy metrics. However, they rarely describe the reliability of the prediction. In this paper, we propose an Uncertainty-Aware Learning (UAL) method to remedy this issue. UAL aims at providing reliability-aware predictions by considering data uncertainty and model uncertainty simultaneously. Data uncertainty captures the "noise" inherent in the sample, while model uncertainty depicts the model's confidence in the sample's prediction. Specifically, in UAL, (1) we propose a sampling-free data uncertainty learning method to adaptively assign weights to different samples during training, down-weighting the lowquality ambiguous samples. (2) we leverage the Bayesian framework to model the model uncertainty by assuming the parameters of the network follow a Bernoulli distribution. (3) the data uncertainty and the model uncertainty are jointly learned in a unified network, and they serve as two fundamental criteria for the reliability assessment: if a probe is highquality (low data uncertainty) and the model is confident in the prediction of the probe (low model uncertainty), the final ranking will be assessed as reliable. Experiments under the risk-controlled settings and the multi-query settings show the proposed reliability assessment is effective. Our method also shows superior performance on three challenging benchmarks under the vanilla single query settings. The code is available at: https://github.com/dcp15/UAL