Picking Deep Filter Responses for Fine-Grained Image Recognition

By Xiaopeng Zhang, H. Xiong, Wen-gang Zhou, Weiyao Lin, Qi Tian. 2016

339 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition

How researchers see this paper

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

In Aerospace Engineering

1. Jiebo Luo
2. Bernt Schiele
3. Yu Qiao
4. Errui Ding
5. Ze Yang
6. Shuicheng Yan
7. Dacheng Tao
8. Qiang Peng

In Computer Vision and Pattern Recognition

1. Dacheng Tao
2. Luc Van Gool
3. Xiaogang Wang
4. Trevor Darrell
5. Yi Yang
6. Yu Qiao
7. Ming-Hsuan Yang
8. Jiashi Feng

Citations in the corpus

• Large Scale Fine-Grained Categorization and Domain-Specific Transfer Learning (2018, neutral): “[68] 84.”
• Attention-Based Dropout Layer for Weakly Supervised Object Localization (2019, background): “In other words, the background of this dataset is nearly independent of classes, thus the background is not a discriminative region [61].”
• Look Closer to See Better: Recurrent Attention Convolutional Neural Network for Fine-Grained Image Recognition (2017, methodology): “Compared with unsupervised methods PDFR [34] without additional Fish- er Vector learning, we can obtain a relative accuracy gain with 3.3% by our full model RA-CNN (scale 1+2+3).”
• Diversified Visual Attention Networks for Fine-Grained Object Classification (2016, methodology): “Different from our attention model which implicitly finds the discriminative parts of the birds, NAC [46] and PD [47] explicitly use specific channels of a CNN as part detectors.”
• WildFish: A Large Benchmark for Fish Recognition in the Wild (2018, methodology): “Over the past years, fine-grained recognition has been mainly driven by deep learning [12, 26, 48].”
• Looking for the Devil in the Details: Learning Trilinear Attention Sampling Network for Fine-Grained Image Recognition (2019, background): “As shown in previous work [24, 39], each channel of the convolutional features corresponds to a visual pattern, however, such feature maps cannot act as attention maps due to the lack of consistency and robustness [32, 40].”
• Mask-CNN: Localizing parts and selecting descriptors for fine-grained bird species categorization (2018, methodology): “On the other hand, for the methods using only image-level labels, such as PDFS [35], Spatial Transformer CNN [26] and Bilinear [27], they are two outstanding fine-grained methods using only the image-level supervisions.”
• Fine-Grained Representation Learning and Recognition by Exploiting Hierarchical Semantic Embedding (2018, methodology): “√ 82.8 CB-CNN w/ bbox [9] √ 84.6 FCAN w/ bbox [28] √ 84.7 B-CNN w/ bbox [25] √ 85.1 AGAL w/ bbox [27] √ 85.5 TLAN [43] 77.9 DVAN [49] 79.0 MG-CNN [40] 81.7 B-CNN w/o bbox [25] 84.1 ST-CNN [17] 84.1 FCAN w/o bbox [28] 84.3 PDFR [48] 84.5 CB-CNN w/o bbox [9] 85.0 RA-CNN [8] 85.3 AGAL w/o bbox [27] 85.”
• Learning Rich Part Hierarchies With Progressive Attention Networks for Fine-Grained Image Recognition (2020, methodology): “, stripe) [13], [19], it is usually difficult to express rich and robust part information by a single channel.”
• Attribute-Aware Attention Model for Fine-grained Representation Learning (2018, methodology): “But those methods heavily rely on manually predefined parts, so some methods proposed to automatically discover critical parts in weak supervised way [10, 49].”
• Object-Part Attention Model for Fine-Grained Image Classification (2017, methodology): “Most existing weakly supervised methods [7], [14] ignore the spatial relationships between the object and its parts as”
• Hierarchical Bilinear Pooling for Fine-Grained Visual Recognition (2018, methodology): “Motivated by the observation that local parts of object usually act a role of importance in differentiating subcategories, many methods [35,2,26,36] for finegrained classification were developed by exploiting the parts, namely part-based ar X iv :1 80 7.”
• Learning Multi-attention Convolutional Neural Network for Fine-Grained Image Recognition (2017, methodology): “The framework are typically composed of two independent steps: 1) part localization by training from positive/nagetive image patches [35] or pinpointing from pre-trained feature channels [25], and 2) fine-grained feature learning by selective pooling [31] or dense encoding from feature maps [17].”