| 1. |
- Beh, Jing Chong, et al.
(author)
-
CyEDA : CYCLE OBJECT EDGE CONSISTENCY DOMAIN ADAPTATION
- 2022
-
In: Proceedings - International Conference on Image Processing, ICIP. - 1522-4880. ; , s. 2986-2990
-
Conference paper (peer-reviewed)abstract
- Despite the advent of domain adaptation methods, most of them still struggle in preserving the instance level details of images when performing global level translation. While there are instance level translation methods that can retain the instance level details well, most of them require either pre-train object detection/segmentation network and annotation labels. In this work, we propose a novel method namely CyEDA to perform global level domain adaptation that taking care of image contents without any pre-train networks integration or annotation labels. That is, we introduce masking and cycle-object edge consistency loss which exploit the preservation of image objects. We show that our approach is able to outperform other SOTAs in terms of image quality and FID score in both BDD100K and GTA datasets.
|
|
| 2. |
- Hsu, Pohao, et al.
(author)
-
Extremely Low-light Image Enhancement with Scene Text Restoration
- 2022
-
In: Proceedings - International Conference on Pattern Recognition. - 1051-4651. - 9781665490627 ; 2022-August, s. 317-323
-
Conference paper (peer-reviewed)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.
|
|
| 3. |
- Le, Huu, 1988, et al.
(author)
-
AdaSTE: An Adaptive Straight-Through Estimator to Train Binary Neural Networks
- 2022
-
In: Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition. - 1063-6919. ; 2022-June, s. 460-469
-
Conference paper (peer-reviewed)abstract
- We propose a new algorithm for training deep neural networks (DNNs) with binary weights. In particular, we first cast the problem of training binary neural networks (BiNNs) as a bilevel optimization instance and subsequently construct flexible relaxations of this bilevel program. The resulting training method shares its algorithmic simplicity with several existing approaches to train BiNNs, in particular with the straight-through gradient estimator successfully employed in BinaryConnect and subsequent methods. In fact, our proposed method can be interpreted as an adaptive variant of the original straight-through estimator that conditionally (but not always) acts like a linear mapping in the backward pass of error propagation. Experimental results demonstrate that our new algorithm offers favorable performance compared to existing approaches.
|
|
| 4. |
- Lin, Che-Tsung, 1979, et al.
(author)
-
Cycle-Object Consistency for Image-to-Image Domain Adaptation
- 2023
-
In: Pattern Recognition. - : Elsevier BV. - 0031-3203. ; 138
-
Journal article (peer-reviewed)abstract
- Recent advances in generative adversarial networks (GANs) have been proven effective in performing domain adaptation for object detectors through data augmentation. While GANs are exceptionally successful, those methods that can preserve objects well in the image-to-image translation task usually require an auxiliary task, such as semantic segmentation to prevent the image content from being too distorted. However, pixel-level annotations are difficult to obtain in practice. Alternatively, instance-aware image-translation model treats object instances and background separately. Yet, it requires object detectors at test time, assuming that off-the-shelf detectors work well in both domains. In this work, we present AugGAN-Det, which introduces Cycle-object Consistency (CoCo) loss to generate instance-aware translated images across complex domains. The object detector of the target domain is directly leveraged in generator training and guides the preserved objects in the translated images to carry target-domain appearances. Compared to previous models, which e.g., require pixel-level semantic segmentation to force the latent distribution to be object-preserving, this work only needs bounding box annotations which are significantly easier to acquire. Next, as to the instance-aware GAN models, our model, AugGAN-Det, internalizes global and object style-transfer without explicitly aligning the instance features. Most importantly, a detector is not required at test time. Experimental results demonstrate that our model outperforms recent object-preserving and instance-level models and achieves state-of-the-art detection accuracy and visual perceptual quality.
|
|
| 5. |
- Lin, Che-Tsung, 1979, et al.
(author)
-
Data Augmentation via Neural-Style-Transfer for Driver Distraction Recognition
- 2022
-
In: The 8th International Conference on Driver Distraction and Inattention.
-
Conference paper (peer-reviewed)abstract
- According to the National Highway Traffic Safety Administration, 3142 people were killed in motor vehicle crashes involving distracted drivers in 2019. Naturalistic driving datasets (NDD) have been widely used to study distracting activities while driving, with the aim of improving road safety. However, the time required to annotate videos to identify distracting activities is a major issue for research using NDD. Although full automation of the annotation process is not possible, the use of image classifiers is a way forward to hasten the classification of distractions and therefore the analysis of NDD. This paper presents the results obtained by applying image classifier to the publicly available Distracted Driver Dataset (DDD) and a sample of frames extracted from the EuroFOT and DriveC2X dataset. The results show that using ResNet-50 pretrained on Imagenet and Stylized Imagenet produces the highest accuracy on both DDD and our EuroFOT and DriveC2X datasets. The accuracy of the image classifier will now be tested on a different sample of the Swedish EuroFOT dataset, before using the image classifier for detecting distracting activities in other NDD. The faster identification of distracting activities will considerably hasten the future analyses of NDD.
|
|
| 6. |
- Liu, Xixi, 1995, et al.
(author)
-
Effortless Training of Joint Energy-Based Models with Sliced Score Matching
- 2022
-
In: Proceedings - International Conference on Pattern Recognition. - 1051-4651. - 9781665490627 ; 2022-August, s. 2643-2649
-
Conference paper (peer-reviewed)abstract
- Standard discriminative classifiers can be upgraded to joint energy-based models (JEMs) by combining the classification loss with a log-evidence loss. Hence, such models intrinsically allow detection of out-of-distribution (OOD) samples, and empirically also provide better-calibrated posteriors, i.e., prediction uncertainties. However, the training procedure suggested for JEMs (using stochastic gradient Langevin dynamics---or SGLD---to maximize the evidence) is reported to be brittle. In this work, we propose to utilize score matching---in particular sliced score matching---to obtain a stable training method for JEMs. We observe empirically that the combination of score matching with the standard classification loss leads to improved OOD detection and better-calibrated classifiers for otherwise identical DNN architectures. Additionally, we also analyze the impact of replacing the regular soft-max layer for classification with a gated soft-max one in order to improve the intrinsic transformation invariance and generalization ability.
|
|
| 7. |
- Liu, Xixi, 1995, et al.
(author)
-
Joint Energy-based Model for Deep Probabilistic Regression
- 2022
-
In: Proceedings - International Conference on Pattern Recognition. - 1051-4651. - 9781665490627 ; 2022-August, s. 2693-2699
-
Conference paper (peer-reviewed)abstract
- It is desirable that a deep neural network trained on a regression task does not only achieve high prediction accuracy, but its prediction posteriors are also well-calibrated, especially in safety-critical settings. Recently, energy-based models specifically to enrich regression posteriors have been proposed and achieve state-of-art results in object detection tasks. However, applying these models at prediction time is not straightforward as the resulting inference methods require to minimize an underlying energy function. Furthermore, these methods empirically do not provide accurate prediction uncertainties. Inspired by recent joint energy-based models for classification, in this work we propose to utilize a joint energy model for regression tasks and describe architectural differences needed in this setting. Within this frame-work, we apply our methods to three computer vision regression tasks. We demonstrate that joint energy-based models for deep probabilistic regression improve the calibration property, do not require expensive inference, and yield competitive accuracy in terms of the mean absolute error (MAE).
|
|
| 8. |
- Mao, Yadong, et al.
(author)
-
Decentralized Training of 3D Lane Detection with Automatic Labeling Using HD Maps
- 2023
-
In: IEEE Vehicular Technology Conference. - 1550-2252. ; 2023-June
-
Conference paper (peer-reviewed)abstract
- To have competent 3D lane detection for real-world driving, a massive amount of data from all over the world is needed, but data collection and manual annotation are costly and time-consuming. The diversity of data collected by developmental cars might still be limited compared to the data collected by a large fleet of customer cars. Federated learning enables training models on edge without transferring data out of devices. However, training supervised learning tasks at the edge is directly tied to having access to high-quality labels, which is limited at the edge. In this paper, we propose a fully automatic method to generate 3D lane labels at the edge using a pre-recorded HD map to enable the federated training of the 3D lane detection model. As a reference, a semi-automatic method is applied for creating a 3D-lane dataset used as ground truth. Our experimental results show that the model can achieve comparable performance when training on the same dataset in both a centralized and a decentralized manner. And the models trained on semi-automatic labeled datasets slightly outperform those trained on fully-automatically labeled datasets. This study shows that a well-performing 3D lane detection model can be trained in a supervised and fully decentralized manner, and most importantly, data privacy at the edge is guaranteed.
|
|
| 9. |
- Nah, Wan Jun, et al.
(author)
-
Rethinking Long-Tailed Visual Recognition with Dynamic Probability Smoothing and Frequency Weighted Focusing
- 2023
-
In: Proceedings - International Conference on Image Processing, ICIP. - 1522-4880. ; , s. 435-439
-
Conference paper (peer-reviewed)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}
|
|
| 10. |
- Ng, Chun Chet, et al.
(author)
-
When IC meets text: Towards a rich annotated integrated circuit text dataset
- 2024
-
In: Pattern Recognition. - 0031-3203. ; 147
-
Journal article (peer-reviewed)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}.
|
|
