Ichiro Ide
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Ichiro Ide has a total of 69 co-authors. Here are the top co-authors by number of shared publications:
- Takahiro Komamizu
Nagoya University, Nagoya, Japan
31 shared publications - Yasutomo Kawanishi
Guardian Robot Project R-IH, RIKEN
26 shared publications - Marc A. Kastner
Hiroshima City University, Hiroshima, Japan
18 shared publications - Takatsugu Hirayama
Nagoya University
16 shared publications - Daisuke Deguchi
Nagoya University, Nagoya, Japan
10 shared publications - Chihaya Matsuhira
Nagoya University, Nagoya, Japan
8 shared publications
Latest Publications
- 15 November 2026
- Published by Elsevier in Neurocomputing
A high-order relation enhanced framework for balancing healthiness and preference in meal recommendation
- 15 November 2026
- Published by Elsevier in Neurocomputing
A high-order relation enhanced framework for balancing healthiness and preference in meal recommendation
- Open Access
- 7 September 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Multimedia Computing, Communications, and Applications
- Vol. 22 (9) , 1-22
Context-aware and View-consistent Learning for Multi-view Action Recognition
The increasing deployment of multi-sensor systems in smart homes, surveillance, and assistive robotics has intensified research on multi-view action recognition. While existing research is effective under fully overlapping sensor configurations, real-world systems typically exhibit partially overlapping views, where actions are visible only in a subset of sensors. This non-uniform visibility, coupled with severe occlusions and inconsistent view coverage across sensors, poses a major challenge for reliable action recognition. Moreover, most current methods focus on implicit feature fusion and overlook explicit reasoning about contextual cues and prediction-level consistency across views, leading to degraded performance under occlusion and misalignment. To address these challenges, we propose Context-aware and View-consistent learning (CoVi) , a unified method that enhances multi-view action recognition by jointly modeling contextual dependencies and cross-view coherence. CoVi introduces two plug-and-play modules: Context-aware and View-aware modules. The former adaptively integrates local human-centric and global scene-level information through gated fusion, enabling the model to emphasize informative regions and suppress background noise for fine-grained contextual reasoning. The latter enforces prediction alignment across co-visible views using confidence-weighted Jensen–Shannon divergence, ensuring consistent learning under view disparity and partial visibility. Extensive experiments conducted on two real-world home environments from the MultiSensor-Home dataset and the MM-Office dataset demonstrate that CoVi consistently outperforms state-of-the-art baselines in both uni-modal and multi-modal settings, validating its effectiveness and generalizability across diverse sensor configurations and scene types.
- 7 September 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Multimedia Computing, Communications, and Applications
- Vol. 22 (9) , 1-22
Open AccessContext-aware and View-consistent Learning for Multi-view Action Recognition
The increasing deployment of multi-sensor systems in smart homes, surveillance, and assistive robotics has intensified research on multi-view action recognition. While existing research is effective under fully overlapping sensor configurations, real-world systems typically exhibit partially overlapping views, where actions are visible only in a subset of sensors. This non-uniform visibility, coupled with severe occlusions and inconsistent view coverage across sensors, poses a major challenge for reliable action recognition. Moreover, most current methods focus on implicit feature fusion and overlook explicit reasoning about contextual cues and prediction-level consistency across views, leading to degraded performance under occlusion and misalignment. To address these challenges, we propose Context-aware and View-consistent learning (CoVi) , a unified method that enhances multi-view action recognition by jointly modeling contextual dependencies and cross-view coherence. CoVi introduces two plug-and-play modules: Context-aware and View-aware modules. The former adaptively integrates local human-centric and global scene-level information through gated fusion, enabling the model to emphasize informative regions and suppress background noise for fine-grained contextual reasoning. The latter enforces prediction alignment across co-visible views using confidence-weighted Jensen–Shannon divergence, ensuring consistent learning under view disparity and partial visibility. Extensive experiments conducted on two real-world home environments from the MultiSensor-Home dataset and the MM-Office dataset demonstrate that CoVi consistently outperforms state-of-the-art baselines in both uni-modal and multi-modal settings, validating its effectiveness and generalizability across diverse sensor configurations and scene types.
- 1 September 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on AgriFood Electronics
- Vol. 4 (2) , 733-747
Leveraging Label-Free and Gradient-Based Pruning for Efficient Neural Networks in Maize Disease Detection
- 1 September 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on AgriFood Electronics
- Vol. 4 (2) , 733-747
Leveraging Label-Free and Gradient-Based Pruning for Efficient Neural Networks in Maize Disease Detection
- 25 June 2026
- Published by Springer Nature
- p. 411-423
Facet-Level Persona Control by Trait-Activated Routing with Contrastive SAE for Role-Playing LLMs
- 25 June 2026
- Published by Springer Nature
- p. 411-423
Facet-Level Persona Control by Trait-Activated Routing with Contrastive SAE for Role-Playing LLMs
- Open Access
- 21 April 2026
- Published by Elsevier in Pattern Recognition
- Vol. 179
MultiSensor-Home: Multi-modal multi-view dataset and benchmarks for action recognition in home environments
- 21 April 2026
- Published by Elsevier in Pattern Recognition
- Vol. 179
Open AccessMultiSensor-Home: Multi-modal multi-view dataset and benchmarks for action recognition in home environments
- Open Access
- 1 February 2026
- Published by Institute of Electronics, Information and Communications Engineers (IEICE) in IEICE Transactions on Information and Systems
- Vol. E109.D (2) , 284-287
Analysis and Prediction of Attractive Fonts on Title-overlaid Food Images
Title-overlaid images are useful as thumbnails on social media, where users prefer concise information to share and watch contents. Focusing on food contents, we aim to support creation of attractive title-overlaid food images to attract viewers' attentions. This paper first analyzes the effect of font styles of the title on the attractiveness of title-overlaid images via preference experiments, and creates a dataset. Next, we design a prototype model of attractive font selection for a food image and its title. Its effectiveness is demonstrated through experiments on the created dataset.
- 1 February 2026
- Published by Institute of Electronics, Information and Communications Engineers (IEICE) in IEICE Transactions on Information and Systems
- Vol. E109.D (2) , 284-287
Open AccessAnalysis and Prediction of Attractive Fonts on Title-overlaid Food Images
Title-overlaid images are useful as thumbnails on social media, where users prefer concise information to share and watch contents. Focusing on food contents, we aim to support creation of attractive title-overlaid food images to attract viewers' attentions. This paper first analyzes the effect of font styles of the title on the attractiveness of title-overlaid images via preference experiments, and creates a dataset. Next, we design a prototype model of attractive font selection for a food image and its title. Its effectiveness is demonstrated through experiments on the created dataset.
- Open Access
- 13 January 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Multimedia Computing, Communications, and Applications
- Vol. 22 (1) , 1-23
Hierarchical Global–Local Fusion for One-stage Open-vocabulary Temporal Action Detection
Open-vocabulary Temporal Action Detection (Open-vocab TAD) extends the detection scope of Closed-vocabulary Temporal Action Detection (Closed-vocab TAD) to unseen action classes specified by vocabularies not included in the training data, within untrimmed video. Typical Open-vocab TAD methods adopt a two-stage approach that first proposes candidate action intervals and then identifies those actions. However, errors in the first stage can affect the subsequent stage and the final detection results. Moreover, conventional methods for temporal context analyses tend to focus solely on either global or local context. Focusing solely on the global context can lead to lack of momentary detail, making it difficult to distinguish one action from another. Conversely, focusing only on the local context makes it challenging to determine the start and end timings of action intervals. To address these challenges, we introduce a one-stage approach named Hierarchical Open-vocab TAD (HOTAD), consisting of two branches: Temporal Context Analysis (TCA) and Video–Text Alignment (VTA). The former utilizes Hierarchical Encoder (HE) to fuse global and local temporal features, enabling a comprehensive capture of temporal actions, while the latter branch exploits the synergy between visual and textual modalities for precisely detecting unseen actions in the Open-vocab setting. Experiments and in-depth analysis using the widely recognized datasets THUMOS14 and ActivityNet-1.3 are performed to show the effectiveness of HOTAD. The results highlight remarkable accuracy in detecting a wide range of unseen actions. Furthermore, HOTAD significantly reduces wrong labels and localizes action instances with high precision, showcasing its robustness in complex and dynamic video settings.
- 13 January 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Multimedia Computing, Communications, and Applications
- Vol. 22 (1) , 1-23
Open AccessHierarchical Global–Local Fusion for One-stage Open-vocabulary Temporal Action Detection
Open-vocabulary Temporal Action Detection (Open-vocab TAD) extends the detection scope of Closed-vocabulary Temporal Action Detection (Closed-vocab TAD) to unseen action classes specified by vocabularies not included in the training data, within untrimmed video. Typical Open-vocab TAD methods adopt a two-stage approach that first proposes candidate action intervals and then identifies those actions. However, errors in the first stage can affect the subsequent stage and the final detection results. Moreover, conventional methods for temporal context analyses tend to focus solely on either global or local context. Focusing solely on the global context can lead to lack of momentary detail, making it difficult to distinguish one action from another. Conversely, focusing only on the local context makes it challenging to determine the start and end timings of action intervals. To address these challenges, we introduce a one-stage approach named Hierarchical Open-vocab TAD (HOTAD), consisting of two branches: Temporal Context Analysis (TCA) and Video–Text Alignment (VTA). The former utilizes Hierarchical Encoder (HE) to fuse global and local temporal features, enabling a comprehensive capture of temporal actions, while the latter branch exploits the synergy between visual and textual modalities for precisely detecting unseen actions in the Open-vocab setting. Experiments and in-depth analysis using the widely recognized datasets THUMOS14 and ActivityNet-1.3 are performed to show the effectiveness of HOTAD. The results highlight remarkable accuracy in detecting a wide range of unseen actions. Furthermore, HOTAD significantly reduces wrong labels and localizes action instances with high precision, showcasing its robustness in complex and dynamic video settings.
- Open Access
- 6 December 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Q-Adapter: Visual Query Adapter for Extracting Textually-related Features in Video Captioning
- 6 December 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Open AccessQ-Adapter: Visual Query Adapter for Extracting Textually-related Features in Video Captioning
- Open Access
Small Object Detection for Birds with Swin Transformer
Object detection is the task of detecting objects in an image. In this task, the detection of small objects is particularly difficult. Other than the small size, it is also accompanied by difficulties due to blur, occlusion, and so on. Current small object detection methods are tailored to small and dense situations, such as pedestrians in a crowd or far objects in remote sensing scenarios. However, when the target object is small and sparse, there is a lack of objects available for training, making it more difficult to learn effective features. In this paper, we propose a specialized method for detecting a specific category of small objects; birds. Particularly, we improve the features learned by the neck; the sub-network between the backbone and the prediction head, to learn more effective features with a hierarchical design. We employ Swin Transformer to upsample the image features. Moreover, we change the shifted window size for adapting to small objects. Experiments show that the proposed Swin Transformer-based neck combined with CenterNet can lead to good performance by changing the window sizes. We further find that smaller window sizes (default 2) benefit mAPs for small object detection.
Open AccessSmall Object Detection for Birds with Swin Transformer
Object detection is the task of detecting objects in an image. In this task, the detection of small objects is particularly difficult. Other than the small size, it is also accompanied by difficulties due to blur, occlusion, and so on. Current small object detection methods are tailored to small and dense situations, such as pedestrians in a crowd or far objects in remote sensing scenarios. However, when the target object is small and sparse, there is a lack of objects available for training, making it more difficult to learn effective features. In this paper, we propose a specialized method for detecting a specific category of small objects; birds. Particularly, we improve the features learned by the neck; the sub-network between the backbone and the prediction head, to learn more effective features with a hierarchical design. We employ Swin Transformer to upsample the image features. Moreover, we change the shifted window size for adapting to small objects. Experiments show that the proposed Swin Transformer-based neck combined with CenterNet can lead to good performance by changing the window sizes. We further find that smaller window sizes (default 2) benefit mAPs for small object detection.
- Open Access
- 10 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 199615-199629
Semantic Alignment on Action for Image Captioning
Image captioning is a popular task in vision and language processing, which aims to generate textual descriptions for images. Previously, it simply used image and text as input with self-attention to capture global dependencies. Recent research further uses objects detected from the input image, so-called object tags, as anchor points to ease alignment between image and text with the attention mechanism. However, they only consider object information in images, while neglecting the actions and object interactions that also appear in the image, which causes actions not caught properly in image captioning. To tackle this previously underrepresented dimension of the semantic alignment, we take account of actions on the semantic level. Specifically, our work focuses on human actions and interactions, which ensures that more salient parts of the image get captioned. We introduce a new type of tag, called action tag, to anchor the action information. First, we provide a method for obtaining such action tags using an action detection model which predicts actions in the image. Next, we leverage these action tags into the captioning model. Experimental results indicate that the proposed action tags can help learn action semantics and catch the salient actions leading to perceived improvements in common performance. Experimental results on MS-COCO Karpathy test split show that the proposed model achieves good scores in BLEU-4 and CIDEr metrics, using action tags as anchors. Furthermore, the number of action tags (no more than 5) is smaller than that of object tags (commonly more than 20), which means there is a potential to reduce FLOPs by reducing the total sequence length. It indicates the potential for efficient reasoning and may be applied to daily activity scenes in the future.
- 10 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 199615-199629
Open AccessSemantic Alignment on Action for Image Captioning
Image captioning is a popular task in vision and language processing, which aims to generate textual descriptions for images. Previously, it simply used image and text as input with self-attention to capture global dependencies. Recent research further uses objects detected from the input image, so-called object tags, as anchor points to ease alignment between image and text with the attention mechanism. However, they only consider object information in images, while neglecting the actions and object interactions that also appear in the image, which causes actions not caught properly in image captioning. To tackle this previously underrepresented dimension of the semantic alignment, we take account of actions on the semantic level. Specifically, our work focuses on human actions and interactions, which ensures that more salient parts of the image get captioned. We introduce a new type of tag, called action tag, to anchor the action information. First, we provide a method for obtaining such action tags using an action detection model which predicts actions in the image. Next, we leverage these action tags into the captioning model. Experimental results indicate that the proposed action tags can help learn action semantics and catch the salient actions leading to perceived improvements in common performance. Experimental results on MS-COCO Karpathy test split show that the proposed model achieves good scores in BLEU-4 and CIDEr metrics, using action tags as anchors. Furthermore, the number of action tags (no more than 5) is smaller than that of object tags (commonly more than 20), which means there is a potential to reduce FLOPs by reducing the total sequence length. It indicates the potential for efficient reasoning and may be applied to daily activity scenes in the future.