[보관함:] AI in Signal division

전기및전자공학부 윤영규 교수 연구팀 생체 형광신호 고정밀 측정을 가능하게 하는 AI 영상 분석 기술 SUPPORT 개발

 

[윤영규 교수 연구팀 생체 형광신호 고정밀 측정을 가능하게 하는 AI 영상 분석 기술 SUPPORT 개발]

1특정 생체 신호의 변화에 비례하여 빛(형광)의 밝기가 변화

 

그림 2. SUPPORT를 활용한 초정밀 신경세포 전압 측정: (상) 원 형광 이미지에서는 낮은 신호대잡음비로 인해 신경세포의 활동전위 관찰이 불가능. (하) SUPPORT를 이용해 신호대잡음비를 높이면, 각 신경세포의 활동전위를 정밀하게 관찰할 수 있음. 

 

 

그림 3. SUPPORT를 활용한 생쥐의 생체 귀 조직 형광 이미지 개선: (좌) 원 형광 이미지에서는 낮은 신호대잡음비로 인해 조직의 세부 구조 관찰이 불가능. (우) SUPPORT를 이용해 신호대잡음비를 높이면 세부 구조와 빠르게 이동하는 적혈구를 관찰할 수 있음.

 

 

그림 4. SUPPORT를 활용한 생쥐의 생체 근조직 형광 이미지 개선: (좌) 원 형광 이미지에서는 낮은 신호대잡음비로 인해 조직의 세부 구조 관찰이 불가능. (우) SUPPORT를 이용해 신호대잡음비를 높이면 근섬유의 세부 구조 및 빠르게 이동하는 적혈구를 관찰할 수 있음.

 

 

The precipitation video datasets have distinctive meteorological patterns where a mass of fluid moves in a particular direction across the entire frames, and each local area of the fluid has an individual life cycle from initiation to maturation to decay. This paper proposes a novel transformer-based model for precipitation nowcasting that can extract global and local dynamics within meteorological characteristics. The experimental results show our model achieves state-of-the-art performances on the precipitation nowcasting benchmark.

 

This paper presents a solution to the Weather4cast 2022 Challenge Stage 2. The goal of the challenge is to forecast future high-resolution rainfall events obtained from ground radar using low-resolution multiband satellite images. We suggest a solution that performs data preprocessing appropriate to the challenge and then predicts rainfall movies using a novel RainUNet. RainUNet is a hierarchical U-shaped network with temporal-wise separable block (TS block) using a decoupled large kernel 3D convolution to improve the prediction performance. Various evaluation metrics show that our solution is effective compared to the baseline method. The source codes are available at https://github.com/jinyxp/Weather4cast-2022.

 

In multi-modal action recognition, it is important to consider not only the complementary nature of different modalities but also global action content. In this paper, we propose a novel network, named Modality Mixer (M-Mixer) network, to leverage complementary information across modalities and temporal context of an action for multi-modal action recognition. We also introduce a simple yet effective recurrent unit, called Multi-modal Contextualization Unit (MCU), which is a core component of M-Mixer. Our MCU temporally encodes a sequence of one modality (e.g., RGB) with action content features of other modalities (e.g., depth, IR). This process encourages M-Mixer to exploit global action content and also to supplement complementary information of other modalities. As a result, our proposed method outperforms state-of-the-art methods on NTU RGB+D 60, NTU RGB+D 120, and NW-UCLA datasets. Moreover, we demonstrate the effectiveness of M-Mixer by conducting comprehensive ablation studies.

 

Standard multi-modal models assume the use of the same modalities in training and inference stages. However, in practice, the environment in which multi-modal models operate may not satisfy such assumption. As such, their performances degrade drastically if any modality is missing in the inference stage. We ask: how can we train a model that is robust to missing modalities? This paper seeks a set of good practices for multi-modal action recognition, with a particular interest in circumstances where some modalities are not available at an inference time. First, we study how to effectively regularize the model during training (e.g., data augmentation). Second, we investigate on fusion methods for robustness to missing modalities: we find that transformer-based fusion shows better robustness for missing modality than summation or concatenation. Third, we propose a simple modular network, ActionMAE, which learns missing modality predictive coding by randomly dropping modality features and tries to reconstruct them with the remaining modality features. Coupling these good practices, we build a model that is not only effective in multi-modal action recognition but also robust to modality missing. Our model achieves the state-of-the-arts on multiple benchmarks and maintains competitive performances even in missing modality scenarios.

 

Previous anti-spoofing methods have used either pseudo maps or user-defined labels, and the performance of each approach depends on the accuracy of the third party networks generating pseudo maps and the way in which the users define the labels. In this paper, we propose a liveness score-based regression network for overcoming the dependency on third party networks and users. First, we introduce a new labeling technique, called pseudo-discretized label encoding for generating discretized labels indicating the amount of information related to real images. Secondly, we suggest the expected liveness score based on a regression network for training the difference between the proposed supervision and the expected liveness score. Finally, extensive experiments were conducted on four face anti-spoofing benchmarks to verify our proposed method on both intra-and cross-dataset tests. The experimental results show our approach outperforms previous methods.

 

Deep neural networks are widely known to be susceptible to adversarial examples, which can cause incorrect predictions through subtle input modifications. These adversarial examples tend to be transferable between models, but targeted attacks still have lower attack success rates due to significant variations in decision boundaries. To enhance the transferability of targeted adversarial examples, we propose introducing competition into the optimization process. Our idea is to craft adversarial perturbations in the presence of two new types of competitor noises: adversarial perturbations towards different target classes and friendly perturbations towards the correct class. With these competitors, even if an adversarial example deceives a network to extract specific features leading to the target class, this disturbance can be suppressed by other competitors. Therefore, within this competition, adversarial examples should take different attack strategies by leveraging more diverse features to overwhelm their interference, leading to improving their transferability to different models. Considering the computational complexity, we efficiently simulate various interference from these two types of competitors in feature space by randomly mixing up stored clean features in the model inference and named this method Clean Feature Mixup (CFM). Our extensive experimental results on the ImageNet-Compatible and CIFAR-10 datasets show that the proposed method outperforms the existing baselines with a clear margin. Our code is available at https://github.com/dreamflake/CFM.

 

Online action detection, which aims to identify an ongoing action from a streaming video, is an important subject in real-world applications. For this task, previous methods use recurrent neural networks for modeling temporal relations in an input sequence. However, these methods overlook the fact that the input image sequence includes not only the action of interest but background and irrelevant actions. This would induce recurrent units to accumulate unnecessary information for encoding features on the action of interest. To overcome this problem, we propose a novel recurrent unit, named Information Discrimination Unit (IDU), which explicitly discriminates the information relevancy between an ongoing action and others to decide whether to accumulate the input information. This enables learning more discriminative representations for identifying an ongoing action. In this paper, we further present a new recurrent unit, called Information Integration Unit (IIU), for action anticipation. Our IIU exploits the outputs from IDN as pseudo action labels as well as RGB frames to learn enriched features of observed actions effectively. In experiments on TVSeries and THUMOS-14, the proposed methods outperform state-of-the-art methods by a significant margin in online action detection and action anticipation. Moreover, we demonstrate the effectiveness of the proposed units by conducting comprehensive ablation studies.

Recently, the widespread use of smart devices has invoked more interest in sensor-based applications. Human activity recognition (HAR) with body-worn sensors is an underlying task that aims to recognize a person’s physical activity from on-body sensor readings. In this article, we address the HAR problem that utilizes accelerometers and gyroscopes. While deep-learning-based feature extraction methods are advancing, recent HAR systems with multimodal sensors mostly use data-level fusion, aggregating different sensor signals into one multichannel input. However, they neglect the fact that different sensors capture different physical properties and produce distinct patterns. In this article, we propose a two-stream convolutional neural network (CNN) model that processes the accelerometer and gyroscope signals separately. The modality-specific features are fused in feature-level and jointly used for the recognition task. Furthermore, we introduce a self-supervised learning (SSL) task that pairs the accelerometer and the gyroscope embeddings acquired from the same activity instance. This auxiliary objective allows the feature extractors of our model to communicate during training and exploit complementary information, achieving better representations for HAR. We name our end-to-end multimodal HAR system the Contrastive Accelerometer–Gyroscope Embedding (CAGE) model. CAGE outperforms preceding HAR models in four publicly available benchmarks. The code is available at github.com/quotation2520/CAGE4HAR.

Recently, the widespread use of smart devices has invoked more interest in sensor-based applications. Human activity recognition (HAR) with body-worn sensors is an underlying task that aims to recognize a person’s physical activity from on-body sensor readings. In this article, we address the HAR problem that utilizes accelerometers and gyroscopes. While deep-learning-based feature extraction methods are advancing, recent HAR systems with multimodal sensors mostly use data-level fusion, aggregating different sensor signals into one multichannel input. However, they neglect the fact that different sensors capture different physical properties and produce distinct patterns. In this article, we propose a two-stream convolutional neural network (CNN) model that processes the accelerometer and gyroscope signals separately. The modality-specific features are fused in feature-level and jointly used for the recognition task. Furthermore, we introduce a self-supervised learning (SSL) task that pairs the accelerometer and the gyroscope embeddings acquired from the same activity instance. This auxiliary objective allows the feature extractors of our model to communicate during training and exploit complementary information, achieving better representations for HAR. We name our end-to-end multimodal HAR system the Contrastive Accelerometer–Gyroscope Embedding (CAGE) model. CAGE outperforms preceding HAR models in four publicly available benchmarks. The code is available at github.com/quotation2520/CAGE4HAR.

 

 

-메인 영상: https://youtu.be/JC1_bnTxPiQ 

-쿠키 영상: https://youtu.be/mhUUZVbeDA0