| 1. |
- Hsu, Pohao, et al.
(författare)
-
Extremely Low-light Image Enhancement with Scene Text Restoration
- 2022
-
Ingår i: Proceedings - International Conference on Pattern Recognition. - 1051-4651. - 9781665490627 ; 2022-August, s. 317-323
-
Konferensbidrag (refereegranskat)abstract
- Deep learning based methods have made impressive progress in enhancing extremely low-light images - the image quality of the reconstructed images has generally improved. However, we found out that most of these methods could not sufficiently recover the image details, for instance the texts in the scene. In this paper, a novel image enhancement framework is proposed to specifically restore the scene texts, as well as the overall quality of the image simultaneously under extremely low-light images conditions. Particularly, we employed a selfregularised attention map, an edge map, and a novel text detection loss. The quantitative and qualitative experimental results have shown that the proposed model outperforms stateof-the-art methods in terms of image restoration, text detection, and text spotting on See In the Dark and ICDAR15 datasets.
|
|
| 2. |
- Meskó, Norbert, et al.
(författare)
-
Exploring Attitudes Toward “Sugar Relationships” Across 87 Countries : A Global Perspective on Exchanges of Resources for Sex and Companionship
-
Ingår i: Archives of Sexual Behavior. - 0004-0002.
-
Tidskriftsartikel (refereegranskat)abstract
- The current study investigates attitudes toward one form of sex for resources: the so-called sugar relationships, which often involve exchanges of resources for sex and/or companionship. The present study examined associations among attitudes toward sugar relationships and relevant variables (e.g., sex, sociosexuality, gender inequality, parasitic exposure) in 69,924 participants across 87 countries. Two self-report measures of Acceptance of Sugar Relationships (ASR) developed for younger companion providers (ASR-YWMS) and older resource providers (ASR-OMWS) were translated into 37 languages. We tested cross-sex and cross-linguistic construct equivalence, cross-cultural invariance in sex differences, and the importance of the hypothetical predictors of ASR. Both measures showed adequate psychometric properties in all languages (except the Persian version of ASR-YWMS). Results partially supported our hypotheses and were consistent with previous theoretical considerations and empirical evidence on human mating. For example, at the individual level, sociosexual orientation, traditional gender roles, and pathogen prevalence were significant predictors of both ASR-YWMS and ASR-OMWS. At the country level, gender inequality and parasite stress positively predicted the ASR-YWMS. However, being a woman negatively predicted the ASR-OMWS, but positively predicted the ASR-YWMS. At country-level, ingroup favoritism and parasite stress positively predicted the ASR-OMWS. Furthermore, significant cross-subregional differences were found in the openness to sugar relationships (both ASR-YWMS and ASR-OMWS scores) across subregions. Finally, significant differences were found between ASR-YWMS and ASR-OMWS when compared in each subregion. The ASR-YWMS was significantly higher than the ASR-OMWS in all subregions, except for Northern Africa and Western Asia.
|
|
| 3. |
- Nah, Wan Jun, et al.
(författare)
-
Rethinking Long-Tailed Visual Recognition with Dynamic Probability Smoothing and Frequency Weighted Focusing
- 2023
-
Ingår i: Proceedings - International Conference on Image Processing, ICIP. - 1522-4880. ; , s. 435-439
-
Konferensbidrag (refereegranskat)abstract
- Deep learning models trained on long-tailed (LT) datasets often exhibit bias towards head classes with high frequency. This paper highlights the limitations of existing solutions that combine class- and instance-level re-weighting loss in a naive manner. Specifically, we demonstrate that such solutions result in overfitting the training set, significantly impacting the rare classes. To address this issue, we propose a novel loss function that dynamically reduces the influence of outliers and assigns class-dependent focusing parameters. We also introduce a new long-tailed dataset, ICText-LT, featuring various image qualities and greater realism than artificially sampled datasets. Our method has proven effective, outperforming existing methods through superior quantitative results on CIFAR-LT, Tiny ImageNet-LT, and our new ICText-LT datasets. The source code and new dataset are available at \url{https://github.com/nwjun/FFDS-Loss}
|
|
| 4. |
- Ng, Chun Chet, et al.
(författare)
-
When IC meets text: Towards a rich annotated integrated circuit text dataset
- 2024
-
Ingår i: Pattern Recognition. - 0031-3203. ; 147
-
Tidskriftsartikel (refereegranskat)abstract
- Automated Optical Inspection (AOI) is a process that uses cameras to autonomously scan printed circuit boards for quality control. Text is often printed on chip components, and it is crucial that this text is correctly recognized during AOI, as it contains valuable information. In this paper, we introduce \textit{ICText}, the largest dataset for text detection and recognition on integrated circuits. Uniquely, it includes labels for character quality attributes such as low contrast, blurry, and broken. While loss-reweighting and Curriculum Learning (CL) have been proposed to improve object detector performance by balancing positive and negative samples and gradually training the model from easy to hard samples, these methods have had limited success with one-stage object detectors commonly used in industry. To address this, we propose Attribute-Guided Curriculum Learning (AGCL), which leverages the labeled character quality attributes in \textit{ICText}. Our extensive experiments demonstrate that AGCL can be applied to different detectors in a plug-and-play fashion to achieve higher Average Precision (AP), significantly outperforming existing methods on \textit{ICText} without any additional computational overhead during inference. Furthermore, we show that AGCL is also effective on the generic object detection dataset Pascal VOC. Our code and dataset will be publicly available at \href{https://github.com/chunchet-ng/ICText-AGCL}{https://github.com/chunchet-ng/ICText-AGCL}.
|
|
| 5. |
- Ogunbode, Charles, et al.
(författare)
-
Climate anxiety, wellbeing and pro-environmental action : Correlates of negative emotional responses to climate change in 32 countries
- 2022
-
Ingår i: Journal of Environmental Psychology. - : Academic Press. - 0272-4944 .- 1522-9610. ; 84
-
Tidskriftsartikel (refereegranskat)abstract
- This study explored the correlates of climate anxiety in a diverse range of national contexts. We analysed cross-sectional data gathered in 32 countries (N = 12,246). Our results show that climate anxiety is positively related to rate of exposure to information about climate change impacts, the amount of attention people pay to climate change information, and perceived descriptive norms about emotional responding to climate change. Climate anxiety was also positively linked to pro-environmental behaviours and inversely related to mental wellbeing. Notably, climate anxiety had a significant inverse association with mental wellbeing in 31 out of 32 countries, and with pro-environmental behaviour in 24 countries, it only predicted environmental activism in 12 countries. Our findings highlight contextual boundaries to engagement in environmental action as an antidote to climate anxiety, and the broad international significance of negative climate-related emotions as a plausible threat to wellbeing.
|
|
