Learning a Discriminative Filter Bank Within a CNN for Fine-Grained Recognition

By Yaming Wang, Vlad I. Morariu, Larry S. Davis. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2016

358 citations recorded by the source corpus.

Compared to earlier multistage frameworks using CNN features, recent end-to-end deep approaches for fine-grained recognition essentially enhance the mid-level learning capability of CNNs. Previous approaches achieve this by introducing an auxiliary network to infuse localization information into the main classification network, or a sophisticated feature encoding method to capture higher order feature statistics. We show that mid-level representation learning can be enhanced within the CNN framework, by learning a bank of convolutional filters that capture class-specific discriminative patches without extra part or bounding box annotations. Such a filter bank is well structured, properly initialized and discriminatively learned through a novel asymmetric multi-stream architecture with convolutional filter supervision and a non-random layer initialization. Experimental results show that our approach achieves state-of-the-art on three publicly available fine-grained recognition datasets (CUB-200-2011, Stanford Cars and FGVC-Aircraft). Ablation studies and visualizations are provided to understand our approach.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Ocean Engineering

View Paper: https://www.semanticscholar.org/paper/32a40b045e665db39e120c12338f9f1238b0690b

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Ocean Engineering

Citations in the corpus

The Devil is in the Channels: Mutual-Channel Loss for Fine-Grained Image Classification (2020, background): “[42] claimed that improving mid-level convolutional feature representation can bring significant advantages for part-based fine-grained classification.”
Learning Attentive Pairwise Interaction for Fine-Grained Classification (2020, background): “…been recently proposed by designing complex spatial attention mechanisms (e.g., RA-CNN (Fu, Zheng, and Mei 2017), MA-CNN (Zheng et al. 2017)), learning a bank of discriminative ﬁlters (Wang, Morariu, and Davis 2018), guiding region detection with multi-agent cooperation (Yang et al. 2018), etc.”
An AI-based Visual Attention Model for Vehicle Make and Model Recognition (2020, background): “For example, ResNet contains four layer groups, and three of them generate mid-level and high-level features, which are the main useful features for recognizing an object, since the information carried by the lower layers is very fundamental, and features of higher layers are more class-specific [26”
Fine-Grained Visual Classification via Progressive Multi-Granularity Training of Jigsaw Patches (2020, background): “FT VGG (CVPR18) [31] VGG16 77.”
See Better Before Looking Closer: Weakly Supervised Data Augmentation Network for Fine-Grained Visual Classification (2019, neutral): “9 DFL-CNN [29] 87.”
Fine-Grained Image Analysis With Deep Learning: A Survey (2021, methodology): “Therefore, the fine-grained community attempted to employ these filter outputs as part detectors [76], [77], [78], [79], [80], [81], [82], and thus rely on them to conduct localization-classification fine-grained recognition.”
Multi-branch and Multi-scale Attention Learning for Fine-Grained Visual Categorization (2020, neutral): “3 - DFL-CNN [25] 87.”
Channel Expansion Convolutional Network for Image Classification (2020, background): “DFL-CNN [26] designs an asymmetric multi-branch structure, and branches at different levels learn global and local information respectively.”