Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks160 from public data
- Contrastive Learning for Compact Single Image Dehazing925
This paper proposes a novel contrastive regularization (CR) built upon contrastive learning to exploit both the information of hazy images and clear images as negative and positive samples, respectively and develops a compact dehazing network based on autoencoder-like framework.
- Farewell to Mutual Information: Variational Distillation for Cross-Modal Person Re-Identification172
A new strategy, Variational Self-Distillation (VSD), is presented, which provides a scalable, flexible and analytic solution to essentially fitting the mutual information but without explicitly estimating it, and highlights the need to rethink the way to estimate mutual information.
- Not All Pixels Are Matched99
A dense contrastive learning framework (DCLNet) is proposed, which performs pixel-to-pixel dense alignment acting on the intermediate representations, rather than the final deep feature, reducing the modality gap from a new perspective.
- NTIRE 2022 Challenge on Efficient Super-Resolution: Methods and Results94
The NTIRE 2022 challenge was to super-resolve an input image with a magnification factor of ×4 based on pairs of low and corresponding high resolution images and the aim was to design a network for single image super-resolution that achieved improvement of efficiency measured according to several metrics.
- 91
- Unveiling the Power of CLIP in Unsupervised Visible-Infrared Person Re-Identification85
A new prompt learning paradigm for unsupervised visible-infrared person re-identification (USL-VI-ReID) by taking full advantage of the visual-text representation ability from CLIP is proposed and a new memory-swapping contrastive learning is proposed.
- 79
- 72
- Multi-modal In-Context Learning Makes an Ego-evolving Scene Text Recognizer70
E2 STR demonstrates that a regular-sized model is sufficient to achieve effective ICL capabilities in STR, and exhibits remarkable training-free adaptation in var-ious scenarios and outperforms even the fine-tuned state-of-the-art approaches on public benchmarks.
- 68
- Towards Compact Single Image Super-Resolution via Contrastive Self-distillation61
A novel contrastive self-distillation (CSD) framework to simultaneously compress and accelerate various off-the-shelf SR models to improve the quality of SR images and PSNR/SSIM via explicit knowledge transfer is proposed.
- 55
- 52
- 46
- 43
- GEOcc: Geometrically Enhanced 3D Occupancy Network With Implicit-Explicit Depth Fusion and Contextual Self-Supervision41
GEOcc, a Geometric-Enhanced Occupancy network tailored for vision-only surround-view perception achieves State-of-the-Art performance on the Occ3D-nuScenes dataset with the least image resolution needed and the most weightless image backbone compared with current models, marking an improvement of 3.3% due to the proposed contributions.
- Cross-Domain and Cross-Modal Knowledge Distillation in Domain Adaptation for 3D Semantic Segmentation41
This paper designs the multi-modal style transfer to convert source image and point cloud to target style and proposes a novel model (Dual-Cross) that integrates Cross-Domain knowledge distillation (CDKD) and Cross-Modal Knowledge Distillation (CMKD), which significantly outperforms both uni- modal and cross-modAL state-of-the-art methods.
- Boosting Night-Time Scene Parsing With Learnable Frequency39
This paper proposes to exploit the image frequency distributions for night-time scene parsing with a Learnable Frequency Encoder (LFE) to model the relationship between different frequency coefficients to measure all frequency components dynamically and proposes a Spatial Frequency Fusion module (SFF) that fuses both spatial and frequency information to guide the extraction of spatial context features.
- 38
- Learning Commonality, Divergence and Variety for Unsupervised Visible-Infrared Person Re-identification37
This work proposes a Progressive Contrastive Learning with Hard and Dynamic Prototypes method for USVI-ReID that generates the hard prototype by selecting the sample with the maximum distance from the cluster center and theoretically shows that the hard prototype is used in the contrastive loss to emphasize divergence.
- Image Understands Point Cloud: Weakly Supervised 3D Semantic Segmentation via Association Learning32
This paper designs a dual-branch network equipped with an active labeling strategy to maximize the power of tiny parts of labels and to directly realize 2D-to-3D knowledge transfer and establishes a cross-modal self-training framework, which iterates between parameter updating and pseudolabel estimation.
- 31
- 31
- LiDAR-Camera Panoptic Segmentation via Geometry-Consistent and Semantic-Aware Alignment29
This work proposes LCPS, the first LiDAR-Camera Panoptic Segmentation network that integrates both geometric and semantic fusion information for the entire point cloud.
- 29
- 28
- Self-supervised Exclusive Learning for 3D Segmentation with Cross-Modal Unsupervised Domain Adaptation28
This paper presents self-supervised exclusive learning for cross-modal semantic segmentation under the UDA scenario, which avoids the prohibitive annotation, and proposes a domain-category adversarial learning with category-wise discriminators by constructing the category prototypes for learning domain-invariant features.
- Variational Distillation for Multi-View Learning27
Multi-View Variational Distillation (MV) is presented, tackling the above limitations for generalized multi-view learning and providing key insights into achieving generalized multi-view representations under a rigorous information-theoretic principle.
- 26
- Effective Image Retrieval via Multilinear Multi-Index Fusion25
This paper proposes a new multi-index fusion scheme for image retrieval by formulating this procedure as a multilinear-based optimization problem, so that the complementary information hidden in different indexes can be explored more thoroughly.
- 24
- 20
- Optimization over Disentangled Encoding: Unsupervised Cross-Domain Point Cloud Completion via Occlusion Factor Manipulation20
This paper disentangle partial scans into three (domain, shape, and occlusion) factors to handle the output gap in cross-domain completion and introduces an optimization stage to maximize the consistency between completed shapes and input scans.
- Cross-stream contrastive learning for self-supervised skeleton-based action recognition19
This paper argues that the defect of single-stream contrast and the lack of necessary feature transformation are responsible for easy positives, and therefore proposes a Cross-Stream Contrastive Learning framework for skeleton-based action Representation learning (CSCLR), which exceeds the state-of-the-art methods on a diverse range of evaluation protocols.
- Task-Level Self-Supervision for Cross-Domain Few-Shot Learning18
This work proposes a domain-independent task-level self-supervised (TL-SS) method for cross-domain few-shot learning and designs a generalized task-agnostic test, where intriguing findings highlight the need to re-think the generalization ability of existing few- shot approaches.
- 17
- 17
- Learning Task-Aware Language-Image Representation for Class-Incremental Object Detection17
A learning task-aware language-image representation method that overcomes the existing drawback of directly utilizing the language-image detector for CIOD and achieves state-of-the-art results under the various CIOD settings.
- Real-time augmented reality application in presurgical planning and lesion scalp localization by a smartphone17
The smartphone augmented reality (AR) application provides a new way of experience to observe intracranial anatomy in situ, and it makes surgical planning more intuitive and efficient.
- Improving Cross-Modal Recipe Retrieval with Component-Aware Prompted CLIP Embedding14
A Component-aware Instance-specific Prompt learning (CIP) model that fully exploits the ability of large-scale VLP models to learn the structured recipe information and therefore allows for aligning visual-textual representations without fine-tuning is proposed.
- 12
- ShakingBot: dynamic manipulation for bagging12
The performance of dynamic shaking action compared to the quasi-static manipulation in the bagging task is demonstrated and it is shown that the method generalizes to variations despite the bag’s size, pattern, and color.
- Learn from Concepts: Towards the Purified Memory for Few-shot Learning12
A novel purified memory mechanism is presented that simulates the recognition process of human beings and enables the model to purify the information from semantic labels and progressively learn consistent, stable, and expressive concepts when episodes are trained one by one.
- 11
- High-throughput “read-on-ski” automated imaging and label-free detection system for toxicity screening of compounds using personalised human kidney organoids11
An automated continuous imaging setup with the “read-on-ski” law of control is developed to maximize temporal resolution with minimum culture plate vibration and opens the door for further application of scaled-up screening using organoids in basic research and drug discovery.
- 11
- 10
- Automatic system for high-throughput and high-sensitivity diagnosis of SARS-CoV-210
An automatic system for SARS-CoV-2 diagnosis is designed and built to deliver high specification, high sensitivity, and high throughput with minimal workforce involvement, and could be further improved using a CRISPR-based assay.
- 8
- 7
- 7
- 7
- 6
- 6
- Hierarchical Walking Transformer for Object Re-Identification6
A new hierarchical walking attention is presented, which provides a scalable, flexible, and interpretable sparsification strategy to reduce the complexity from quadratic to linear, and meanwhile evidently boost the performance.
- 6
- 5
- 5
- 5
- 4
- 4
- 4
- 4
- 4
- 3
- PFDepth: Heterogeneous Pinhole-Fisheye Joint Depth Estimation via Distortion-aware Gaussian-Splatted Volumetric Fusion3
This is the first systematic study of heterogeneous pinhole-fisheye depth estimation, offering both technical novelty and valuable empirical insights, and sets a state-of-the-art performance on KITTI-360 and RealHet datasets over current mainstream depth networks.
- 3
- 3
- 3
- 3
- 2
- Wandering and feeling the Scenes: Body-Aware Diffusion for 3D Human Motion Generation2
TheBA-IDM is proposed, which enables fine-grained control of human motion within a scene by leveraging multimodal information, and is designed to operate on de-identified motion data, effectively protecting user privacy throughout the process.
- 2
- 2
- 2
- 1
- 1
- 1
- 1
- Optimal Transport with Arbitrary Prior for Dynamic Resolution Network1
A novel optimal transport dynamic resolution network (OTD-Net) is proposed by establishing an intrinsic connection between resolution assignment and optimal transport problem, and a multi-resolution inference is proposed to ensemble low-resolution predictions.
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Isolation and Integration: A Strong Pre-trained Model-Based Paradigm for Class-Incremental Learning–
This work proposes a brand-new stage-isolation based class incremental learning framework, which leverages parameter-efficient tuning technique to finetune the pre-trained model for each task, thus mitigating information interference and conflicts among tasks.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Man Must Have Travel Stories to Tell:Reading of the Relation between Literature and Reality in China–
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-16. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.