Rethinking Feature Discrimination and Polymerization for Large-scale Recognition

Abstract

Feature matters. How to train a deep network to acquire discriminative features across categories and polymerized features within classes has always been at the core of many computer vision tasks, specially for large-scale recognition systems where test identities are unseen during training and the number of classes could be at million scale. In this paper, we address this problem based on the simple intuition that the cosine distance of features in high-dimensional space should be close enough within one class and far away across categories. To this end, we proposed the congenerous cosine (COCO) algorithm to simultaneously optimize the cosine similarity among data. It inherits the softmax property to make inter-class features discriminative as well as shares the idea of class centroid in metric learning. Unlike previous work where the center is a temporal, statistical variable within one mini-batch during training, the formulated centroid is responsible for clustering inner-class features to enforce them polymerized around the network truncus. COCO is bundled with discriminative training and learned end-to-end with stable convergence. Experiments on five benchmarks have been extensively conducted to verify the effectiveness of our approach on both small-scale classification task and large-scale human recognition problem.

References in corpus (9)

• Large-Margin Softmax Loss for Convolutional Neural Networks
• SphereFace: Deep Hypersphere Embedding for Face Recognition
• Convolutional Pose Machines
• Unsupervised Learning of Visual Representations using Videos
• Dual Deep Network for Visual Tracking
• Multi-Bias Non-linear Activation in Deep Neural Networks
• Beyond Frontal Faces: Improving Person Recognition Using Multiple Cues
• Faceless Person Recognition; Privacy Implications in Social Media
• Learning a Metric Embedding for Face Recognition using the Multibatch Method

Cited by in corpus (25)

• A Survey of Deep Learning-based Object Detection
• Deep Face Recognition: A Survey
• The Devil is in the Channels: Mutual-Channel Loss for Fine-Grained Image Classification
• Learning Imbalanced Datasets with Label-Distribution-Aware Margin Loss
• Learning to Disentangle Scenes for Person Re-identification
• Deep Imbalanced Learning for Face Recognition and Attribute Prediction
• AdaCos: Adaptively Scaling Cosine Logits for Effectively Learning Deep Face Representations
• Region-based Quality Estimation Network for Large-scale Person Re-identification
• Large Margin Few-Shot Learning
• Generalized Inter-class Loss for Gait Recognition
• Feature Intertwiner for Object Detection
• Person Recognition in Personal Photo Collections
• Multi-label Contrastive Predictive Coding
• Prototype Memory for Large-scale Face Representation Learning
• Learning towards Minimum Hyperspherical Energy
• Beyond Trade-off: Accelerate FCN-based Face Detector with Higher Accuracy
• IronMask: Modular Architecture for Protecting Deep Face Template
• Worse WER, but Better BLEU? Leveraging Word Embedding as Intermediate in Multitask End-to-End Speech Translation
• Hierarchy Neighborhood Discriminative Hashing for An Unified View of Single-Label and Multi-Label Image retrieval
• Zoom-RNN: A Novel Method for Person Recognition Using Recurrent Neural Networks
• ReMarNet: Conjoint Relation and Margin Learning for Small-Sample Image Classification
• Fine-Grained Visual Classification via Simultaneously Learning of Multi-regional Multi-grained Features
• Alpha-Net: Architecture, Models, and Applications
• Profile to Frontal Face Recognition in the Wild Using Coupled Conditional GAN
• Rethinking Loss Design for Large-scale 3D Shape Retrieval