허재필 교수 연구실, CVPR 2022 논문 2편 게재 승인
- 인공지능학과
- 조회수2085
- 2022-03-31
비주얼컴퓨팅연구실(지도교수: 허재필)의 논문 2편이 컴퓨터 비전 및 인공지능 분야의 Top-tier 학술대회인 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2022에 게재 승인되었습니다.
논문 #1: “Local Attention Pyramid for Scene Image Generation” (인공지능학과 박사과정 심상헌, 인공지능학과 박사과정 현상익, 인공지능학과 박사과정 배대현)
논문 #2: “Task Discrepancy Maximization for Fine-grained Few-Shot Classification” (인공지능학과 석사과정 이수빈, 인공지능학과 석사과정 문원준)
“Local Attention Pyramid for Scene Image Generation” 에서는 장면 영상 생성(scene image generation)에서 물체 별 품질 불균형 문제(class-wise visual quality imbalance issue)가 발생한다는 것을 확인하였으며, 이를 완화하기 위해서 새로운 네트워크 모듈을 제안하였습니다. 제안하는 모듈은 장면 영상의 특징인 여러 물체가 전체 영상에 산재해 있다는 점에 착안하여, 생성모델이 여러 로컬 영역들(local regions)에 높은 주의 점수(attention score)를 계산하도록 설계하였습니다. 본 연구에서는 제안된 기법을 이미지 생성분야의 여러 선행 연구들에 적용하였으며, 일관되게 이미지 생성 품질이 향상됨을 확인하였습니다.
“Task Discrepancy Maximization for Fine-grained Few-Shot Classification” 에서는 Fine-grained few-shot classification의 성능향상을 위한 모듈(Task Discrepancy Maximization, TDM)을 제시하였습니다. Few-shot classification은 기존에 학습하지 않았던 카테고리를 적은 수의 라벨링 데이터로 학습하는 분야입니다. 제안하는 모듈은 라벨링 된 데이터(Support-set)을 이용하여 각 카테고리별 차별적인 특징을 예측하고 강조하는 서브모듈(Support Attention Module, SAM)과 분류해야 하는 데이터(Query-set)을 이용하여 분류 대상과 관련 있는 특징을 강조하는 서브모듈(Query Attention Module, QAM)로 구성되어 있습니다. 제안된 모듈은 기존 방법들의 성능을 모두 향상시켰으며 Fine-grained few-shot classification 분야에서 가장 높은 성능을 달성하였습니다.
[논문 #1 정보]
Local Attention Pyramid for Scene Image Generation
Sang-Heon Shim, Sangeek Hyun, DaeHyun Bae, and Jae-Pil Heo
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022
Abstract:
In this paper, we first investigate the class-wise visual quality imbalance problem of scene images generated by GANs. The tendency is empirically found that the class-wise visual qualities are highly correlated with the dominance of object classes in the training data in terms of their scales and appearance frequencies. Specifically, the synthesized qualities of small and less frequent object classes tend to be low. To address this, we propose a novel attention module, Local Attention Pyramid (LAP) module tailored for scene image synthesis, that encourages GANs to generate diverse object classes in a high quality by explicit spread of high attention scores to local regions, since objects in scene images are scattered over the entire images. Moreover, our LAP assigns attention scores in a multiple scale to reflect the scale diversity of various objects. The experimental evaluations on three different datasets show consistent improvements in Frechet Inception Distance (FID) and Frechet Segmentation Distance (FSD) over the state-of-the-art baselines. Furthermore, we apply our LAP module to various GANs methods to demonstrate a wide applicability of our LAP module.
[논문 #2 정보]
Task Discrepancy Maximization for Fine-grained Few-Shot Classification
SuBeen Lee, WonJun Moon, and Jae-Pil Heo
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022 [Oral presentation]
Abstract:
Recognizing discriminative details such as eyes and beaks is important for distinguishing fine-grained classes since they have similar overall appearances. In this regard, we introduce Task Discrepancy Maximization (TDM), a simple module for fine-grained few-shot classification. Our objective is to localize the class-wise discriminative regions by highlighting channels encoding distinct information of the class. Specifically, TDM learns task-specific channel weights based on two novel components: Support Attention Module (SAM) and Query Attention Module (QAM). SAM produces a support weight to represent channel-wise discriminative power for each class. Still, since the SAM is basically only based on the labeled support sets, it can be vulnerable to bias toward such support set. Therefore, we propose QAM which complements SAM by yielding a query weight that grants more weight to object-relevant channels for a given query image. By combining these two weights, a class-wise task-specific channel weight is defined. The weights are then applied to produce task-adaptive feature maps more focusing on the discriminative details. Our experiments validate the effectiveness of TDM and its complementary benefits with prior methods in fine-grained few-shot classification.