no code implementations • 22 Apr 2023 • Eunseop Lee, Inhan Kim, Daijin Kim
In addition, SEM obtains class-related feature representations using the classifier weight and focuses on the foreground features for domain adaptation.
2 code implementations • 20 Sep 2022 • Taehun Kim, Kunhee Kim, Joonyeong Lee, Dongmin Cha, Jiho Lee, Daijin Kim
Salient object detection (SOD) has been in the spotlight recently, yet has been studied less for high-resolution (HR) images.
Ranked #1 on RGB Salient Object Detection on ECSSD
1 code implementation • 16 Aug 2022 • Jinhwan Seo, Wonho Bae, Danica J. Sutherland, Junhyug Noh, Daijin Kim
Weakly Supervised Object Detection (WSOD) is a task that detects objects in an image using a model trained only on image-level annotations.
Ranked #1 on Weakly Supervised Object Detection on MS-COCO-2017
1 code implementation • AAAI 2022 • Inhan Kim, Joonyeong Lee, Daijin Kim
The domain-specific features are used to calculate the importance weight of the domain-specific experts, and the disentangled domain-general and dynamic-object features are applied in estimating the control values.
no code implementations • 20 Apr 2022 • Dongmin Cha, Daijin Kim
Deep neural advancements have recently brought remarkable image synthesis performance to the field of image inpainting.
1 code implementation • CVPR 2022 • Kunhee Kim, Sanghun Park, Eunyeong Jeon, Taehun Kim, Daijin Kim
Current image-to-image translations do not control the output domain beyond the classes used during training, nor do they interpolate between different domains well, leading to implausible results.
Image Manipulation Multimodal Unsupervised Image-To-Image Translation +1
no code implementations • 2 Sep 2021 • Eunyeong Jeon, Kunhee Kim, Daijin Kim
Secondly, we introduce a feature-aware loss to provide the generator more direct supervision by employing the feature representation from the self-supervised discriminator.
no code implementations • 25 Aug 2021 • Sanghun Park, Kunhee Kim, Eunseop Lee, Daijin Kim
Object detection has been applied in a wide variety of real world scenarios, so detection algorithms must provide confidence in the results to ensure that appropriate decisions can be made based on their results.
1 code implementation • 6 Jul 2021 • Taehun Kim, Hyemin Lee, Daijin Kim
We construct a modified version of U-Net shape network with additional encoder and decoder and compute a saliency map in each bottom-up stream prediction module and propagate to the next prediction module.
Ranked #8 on Medical Image Segmentation on ETIS-LARIBPOLYPDB
no code implementations • 8 Jun 2021 • Taehun Kim, Jinseong Kim, Daijin Kim
For this reason, we propose Spatial Context Memoization (SpaM), a bypassing branch for spatial context by retaining the input dimension and constantly communicating its spatial context and rich semantic information mutually with the backbone network.
no code implementations • 5 Sep 2019 • Yonghyun Kim, Bong-Nam Kang, Daijin Kim
An image pyramid can extend many object detection algorithms to solve detection on multiple scales.
no code implementations • ICCV 2019 • Bong-Nam Kang, Yonghyun Kim, Bongjin Jun, Daijin Kim
In this paper, we propose a novel face recognition method, called Attentional Feature-pair Relation Network (AFRN), which represents the face by the relevant pairs of local appearance block features with their attention scores.
no code implementations • 15 Nov 2018 • Bong-Nam Kang, Yonghyun Kim, Daijin Kim
We propose a new face recognition method, called a pairwise relational network (PRN), which takes local appearance features around landmark points on the feature map, and captures unique pairwise relations with the same identity and discriminative pairwise relations between different identities.
no code implementations • 13 Nov 2018 • Yonghyun Kim, Taewook Kim, Bong-Nam Kang, Jieun Kim, Daijin Kim
To verify our method, we visualize the activation of the sub-networks according to the boundary contexts and empirically show that the sub-networks contribute more to the related sub-problem in detection.
no code implementations • ECCV 2018 • Yonghyun Kim, Bong-Nam Kang, Daijin Kim
However, due to the lack of scale normalization in CNN-based detection methods, the activated channels in the feature space can be completely different according to a scale and this difference makes it hard for the classifier to learn samples.
no code implementations • ECCV 2018 • Bong-Nam Kang, Yonghyun Kim, Daijin Kim
Because the existence and meaning of pairwise relations should be identity dependent, we add a face identity state feature, which obtains from the long short-term memory (LSTM) units network with the sequential local appearance patches on the feature maps, to the PRN.