Destruction and Construction Learning for Fine-Grained Image Recognition

By Yue Chen, Yalong Bai, Wei Zhang, Tao Mei. Computer Vision and Pattern Recognition, 2019

474 citations recorded by the source corpus.

Delicate feature representation about object parts plays a critical role in fine-grained recognition. For example, experts can even distinguish fine-grained objects relying only on object parts according to professional knowledge. In this paper, we propose a novel "Destruction and Construction Learning" (DCL) method to enhance the difficulty of fine-grained recognition and exercise the classification model to acquire expert knowledge. Besides the standard classification backbone network, another "destruction and construction" stream is introduced to carefully "destruct" and then "reconstruct" the input image, for learning discriminative regions and features. More specifically, for "destruction", we first partition the input image into local regions and then shuffle them by a Region Confusion Mechanism (RCM). To correctly recognize these destructed images, the classification network has to pay more attention to discriminative regions for spotting the differences. To compensate the noises introduced by RCM, an adversarial loss, which distinguishes original images from destructed ones, is applied to reject noisy patterns introduced by RCM. For "construction", a region alignment network, which tries to restore the original spatial layout of local regions, is followed to model the semantic correlation among local regions. By jointly training with parameter sharing, our proposed DCL injects more discriminative local details to the classification network. Experimental results show that our proposed framework achieves state-of-the-art performance on three standard benchmarks. Moreover, our proposed method does not need any external knowledge during training, and there is no computation overhead at inference time except the standard classification network feed-forwarding. Source code: https://github.com/JDAI-CV/DCL.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Radiology, Nuclear Medicine and Imaging

View Paper: https://www.semanticscholar.org/paper/c947c89c4a709ad27fe4590294589196642b0214

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

1. Haibin Ling
2. Ayan Kumar Bhunia
3. Yi-Zhe Song
4. Xiu-Shen Wei
5. Jian Yang
6. Jianxin Wu
7. Jinhui Tang
8. Longyin Wen

In Computer Vision and Pattern Recognition

1. Luc Van Gool
2. Dacheng Tao
3. Ming-Hsuan Yang
4. Yi-Zhe Song
5. Tao A. Xiang
6. Lei Zhang
7. Xiaogang Wang
8. Yi Yang

In Radiology, Nuclear Medicine and Imaging

1. Xiao Bian
2. Yi-Zhe Song
3. Haibin Ling
4. Dawei Du
5. Libo Zhang
6. Shaopeng Liu
7. Xiu-Shen Wei
8. Longyin Wen

Citations in the corpus

• Fine-Grained Visual Classification via Progressive Multi-Granularity Training of Jigsaw Patches (2020, methodology): “Note that, the very recent work of [4] first adopted a jigsaw solver to solve for fine-grained classification.”
• Counterfactual Attention Learning for Fine-Grained Visual Categorization and Re-identification (2021, unknown): “0 DCL [9] 87.”
• Feature Space Augmentation for Long-Tailed Data (2020, background): “For example, in [7], an input image is partitioned into local regions which are shuffled during training.”
• Fine-Grained Image Analysis With Deep Learning: A Survey (2021, background): “DCL [113] performed a “destruction and construction” process to enhance the difficulty of recognition to guide the network to focus on discriminative parts for fine-grained recognition (i.”
• Attention Convolutional Binary Neural Tree for Fine-Grained Visual Categorization (2019, methodology): “…× 86.5 MaxEnt [10] DenseNet-161 × 86.5 HBPASM [40] Resnet-34 × 86.8 DCL [7] VGG-16 × 86.9 KERL w/ HR [6] VGG-16 × 87.0 TASN [53] VGG-19 × 87.1 DFL-CNN [46] ResNet-50 × 87.4 DCL [7] ResNet-50 × 87.8 TASN [53] ResNet-50 × 87. conducted to demonstrate the effectiveness of the proposed method.”