Mask-CNN: Localizing Parts and Selecting Descriptors for Fine-Grained Image Recognition

Abstract

Fine-grained image recognition is a challenging computer vision problem, due to the small inter-class variations caused by highly similar subordinate categories, and the large intra-class variations in poses, scales and rotations. In this paper, we propose a novel end-to-end Mask-CNN model without the fully connected layers for fine-grained recognition. Based on the part annotations of fine-grained images, the proposed model consists of a fully convolutional network to both locate the discriminative parts (e.g., head and torso), and more importantly generate object/part masks for selecting useful and meaningful convolutional descriptors. After that, a four-stream Mask-CNN model is built for aggregating the selected object- and part-level descriptors simultaneously. The proposed Mask-CNN model has the smallest number of parameters, lowest feature dimensionality and highest recognition accuracy when compared with state-of-the-arts fine-grained approaches.

arXiv commentSubmitted to NIPS 2016

References in corpus (3)

• MatConvNet - Convolutional Neural Networks for MATLAB
• Bird Species Categorization Using Pose Normalized Deep Convolutional Nets
• Neural Activation Constellations: Unsupervised Part Model Discovery with Convolutional Networks

Cited by in corpus (15)

• Noisy Softmax: Improving the Generalization Ability of DCNN via Postponing the Early Softmax Saturation
• Multi-branch and Multi-scale Attention Learning for Fine-Grained Visual Categorization
• Fine-Grained Age Estimation in the wild with Attention LSTM Networks
• Learning a Discriminative Filter Bank within a CNN for Fine-grained Recognition
• Weakly Supervised Bilinear Attention Network for Fine-Grained Visual Classification
• Deep Attentional Structured Representation Learning for Visual Recognition
• Weakly-supervised Object Localization for Few-shot Learning and Fine-grained Few-shot Learning
• Learning Class Unique Features in Fine-Grained Visual Classification
• Universal-to-Specific Framework for Complex Action Recognition
• RAMS-Trans: Recurrent Attention Multi-scale Transformer forFine-grained Image Recognition
• DeepIso: A Deep Learning Model for Peptide Feature Detection
• Automatic Surface Area and Volume Prediction on Ellipsoidal Ham using Deep Learning
• Vehicle Re-Identification Based on Complementary Features
• One-Shot Fine-Grained Instance Retrieval
• Exploiting Spline Models for the Training of Fully Connected Layers in Neural Network