Simultaneously Localize, Segment and Rank the Camouflaged Objects

Lv, Yunqiu; Zhang, Jing; Dai, Yuchao; Li, Aixuan; Liu, Bowen; Barnes, Nick; Fan, Deng-Ping

doi:10.1109/cvpr46437.2021.01142

Cited by 168 publications

(129 citation statements)

References 53 publications

Supporting

Mentioning

112

Contrasting

Order By: Relevance

“…In 2021, Mei et al [14] simulated the predation process of animals and proposed PFNet, a camouflaged object segmentation network based on distraction mining. Lv et al [15] proposed a joint learning network that can simultaneously localize, segment and rank camouflaged objects and proposed a new COD dataset called NC4K. However, the design principle and network structure of the existing COD models are relatively complex.…”

Section: Camouflaged Object Detection Based On Deep Learningmentioning

confidence: 99%

MAGNet: A camouflaged object detection network simulating the observation effect of a magnifier

Jiang

Cai

Jiang

et al. 2022

Preprint

View full text Add to dashboard Cite

In recent years, protecting important objects by simulating animal camouflage has been widely used in many fields. Therefore, camouflaged object detection (COD) technology has emerged. COD is more difficult than traditional object detection techniques because of the high degree of fusion of camouflaged objects with the background. In this paper, we strive to more accurately identify camouflaged objects. Inspired by the use of magnifiers to search for hidden objects in pictures, we propose a COD network that simulates the observation effect of a magnifier, called the MAGnifier Network (MAGNet). Specifically, our MAGNet contains two parallel modules, i.e., the ergodic magnification module (EMM) and the attention focus module (AFM). The EMM is designed to mimic the process of a magnifier enlarging an image, and AFM is used to simulate the observation process in which human attention is highly focused on a region. The two sets of output camouflaged object maps are merged to simulate the observation of an object by a magnifier. In addition, a weighted key point area perception loss function, which is more applicable to COD, is designed based on two modules to give higher attention to the camouflaged object. Extensive experiments demonstrate that compared with 14 cutting-edge detection models, MAGNet can achieve the best comprehensive effect on eight evaluation metrics on the public COD dataset, and the segmentation accuracy is significantly improved. We also validate the models' generalization ability on a military camouflaged object dataset constructed in-house. Finally, we experimentally explore some extended applications of COD.

show abstract

Section: Camouflaged Object Detection Based On Deep Learningmentioning

confidence: 99%

MAGNet: A camouflaged object detection network simulating the observation effect of a magnifier

Jiang

Cai

Jiang

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Implementation: We compare the performance of the invertible attention with non-invertible attention in a dense prediction task: camouflaged object detection [39,29,36], which aims to accurately localize the whole scope of the camouflaged objects. Specifically, we adopt the camouflaged object detection network from [27], which takes ResNet50 [15] as the backbone. The model is trained to generate a one channel camouflage map, representing the possibility of each pixel belong to a camouflaged object.…”

Section: Invertible Attention Vs Non-invertible Attention In Discrimi...mentioning

confidence: 99%

“…The model is trained to generate a one channel camouflage map, representing the possibility of each pixel belong to a camouflaged object. We train the model on the training set of the COD10K dataset [10], and evaluate on four public camouflage testing datasets, including CAMO [24], CHAMELEON [40], COD10K testing dataset and NC4K testing dataset [27]. We use four evaluation metrics to evaluate model performance, including Mean Absolute Error, Mean F-measure, Mean E-measure [9] and S-measure [8] denoted as M, F β , E ξ , S α , respectively.…”

Section: Invertible Attention Vs Non-invertible Attention In Discrimi...mentioning

confidence: 99%

“…Quantitative results: To test the effectiveness of both our non-invertible and invertible attention models, we add each attention to each level of the backbone network in [27]. We summarize performance of invertible attention and non-invertible attention in Table 4.…”

Section: Invertible Attention Vs Non-invertible Attention In Discrimi...mentioning

confidence: 99%

“…Dataset: We train the models using the COD10K training dataset [10] of size 4,040, and test them on four test datasets: CAMO [24] (250 images), CHAMELEON [40] (76 images), the COD10K testing dataset (2,026 images), and NC4K [27] (4,121 images).…”

Section: F3 Invertible Attention Vs Non-invertible Attention In Discr...mentioning

confidence: 99%

See 2 more Smart Citations

Invertible Attention

Zha,

Zhong,

Zhang

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Attention has been proved to be an efficient mechanism to capture long-range dependencies. However, so far it has not been deployed in invertible networks. This is due to the fact that in order to make a network invertible, every component within the network needs to be a bijective transformation, but a normal attention block is not. In this paper, we propose invertible attention that can be plugged into existing invertible models. We mathematically and experimentally prove that the invertibility of an attention model can be achieved by carefully constraining its Lipschitz constant. We validate the invertibility of our invertible attention on image reconstruction task with 3 popular datasets: CIFAR-10, SVHN, and CelebA. We also show that our invertible attention achieves similar performance in comparison with normal non-invertible attention on dense prediction tasks.Preprint. Under review.

show abstract

Highly Accurate Dichotomous Image Segmentation

Qin

Dai

et al. 2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Simultaneously Localize, Segment and Rank the Camouflaged Objects

Cited by 168 publications

References 53 publications

MAGNet: A camouflaged object detection network simulating the observation effect of a magnifier

MAGNet: A camouflaged object detection network simulating the observation effect of a magnifier

Invertible Attention

Highly Accurate Dichotomous Image Segmentation

Contact Info

Product

Resources

About