Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileAcademic lineage
Possible advisorsa guess from early papers, not confirmed
- Suggested from co-authorship
- Rynson W. H. LauSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works88 from public data
- Self-supervised Video Representation Learning by Pace Prediction268
This paper addresses the problem of self-supervised video representation learning from a new perspective -- by video pace prediction -- by introducing contrastive learning to push the model towards discriminating different paces by maximizing the agreement on similar video content.
- Self-Supervised Spatio-Temporal Representation Learning for Videos by Predicting Motion and Appearance Statistics226
This paper proposes to learn visual features by regressing both motion and appearance statistics along spatial and temporal dimensions, given only the input video data, and shows that the approach can significantly improve the performance of C3D when applied to video classification tasks.
- Look Deeper into Depth: Monocular Depth Estimation with Semantic Booster and Attention-Driven Loss224
The long tail property is investigated and an attention-driven loss for the network supervision is proposed and a synergy network is proposed to automatically learn the information sharing strategies between the two tasks to better leverage the semantic information for monocular depth estimation.
- Connecting Image Denoising and High-Level Vision Tasks via Deep Learning204
A convolutional neural network in which convolutions are conducted in various spatial resolutions via downsampling and upsampling operations in order to fuse and exploit contextual information on different scales and the benefit of exploiting image semantics simultaneously for image denoising and high-level vision tasks via deep learning is demonstrated.
- When AWGN-Based Denoiser Meets Real Noises176
This paper proposes a novel approach to boost the performance of a real image denoiser which is trained only with synthetic pixel-independent noise data dominated by AWGN, and investigates Pixel-shuffle Down-sampling strategy to adapt the trained model to real noises.
- Affinity Attention Graph Neural Network for Weakly Supervised Semantic Segmentation156
This paper proposes affinity attention graph neural network, in which an affinity attention layer is designed to acquire the short- and long- distance information from soft graph edges to accurately propagate semantic labels from the confident seeds to the unlabeled pixels.
- 108
- 100
- A hand pose tracking benchmark from stereo matching99
A novel stereo-based hand segmentation and depth estimation algorithm specially tailored for hand tracking here and the experiments indicate the effectiveness of the proposed algorithm by demonstrating that its tracking performance is comparable to the use of an active depth sensor under various of challenging scenarios.
- 71
- 69
- Self-Supervised Representation Learning for Ultrasound Video67
This paper proposes a self-supervised learning approach to learn meaningful and transferable representations from medical imaging video without any type of human annotation, and forces the model to address anatomy-aware tasks with free supervision from the data itself.
- 67
- FreMIM: Fourier Transform Meets Masked Image Modeling for Medical Image Segmentation61
A new MIM-based framework named FreMIM is presented for self-supervised pre-training to better accomplish medical image segmentation tasks and incorporates multi-stage supervision to guide the representation learning during the pre-training phase.
- Self-supervised Video Representation Learning by Uncovering Spatio-temporal Statistics61
This paper proposes a novel pretext task to address the self-supervised video representation learning problem, inspired by the observation that human visual system is sensitive to rapidly changing contents in the visual field, and only needs impressions about rough spatial locations to understand the visual contents.
- 59
- 59
- Mining Unseen Classes via Regional Objectness: A Simple Baseline for Incremental Segmentation55
The proposed MicroSeg is based on the assumption that background regions with strong objectness possibly belong to those concepts in the historical or future stages, and first splits the given image into hundreds of segment proposals with a proposal generator, to avoid forgetting old knowledge at the current training stage.
- 53
- Self-Supervised Ultrasound to MRI Fetal Brain Image Synthesis52
The proposed model is end-to-end trainable and self-supervised without any external annotations, and is evaluated quantitatively and qualitatively with comparison to real fetal MR images and other approaches to synthesis, demonstrating its feasibility of synthesising realistic MR images.
- Self-Supervised Contrastive Video-Speech Representation Learning for Ultrasound46
This paper designed a framework to model the correspondence between video and audio without any kind of human annotations, and introduces cross-modal contrastive learning and an affinity-aware self-paced learning scheme to enhance correlation modelling.
- 44
- Task-Driven Webpage Saliency44
An end-to-end learning framework for predicting task-driven visual saliency on webpages using a convolutional neural network to predict where people look at it under different task conditions, achieving state-of-the-art performance on a newly collected benchmark dataset.
- $360+x$: A Panoptic Multi-modal Scene Understanding Dataset43
This dataset offers a panoptic perspective that encapsulate third-person panoramic and front views, as well as egocentric monocular/binocular views with rich modalities including video, multi-channel audio, directional binaural delay, location data and textual scene descriptions within each scene captured, presenting comprehensive observation of the world.
- Frame Interpolation with Consecutive Brownian Bridge Diffusion31
This work proposes consecutive Brownian Bridge diffusion that takes a deterministic initial value as input, resulting in a much smaller cumulative variance of generated latent representations, leaving strong potential for further enhancement in VFI.
- CL-MVSNet: Unsupervised Multi-view Stereo with Dual-level Contrastive Learning31
This paper proposes a new dual-level contrastive learning approach, named CL-MVSNet, which integrates two contrastive branches into an unsupervised MVS framework to construct additional supervisory signals and achieves state-of-the-art performance among all end-to-end unsupervised MVS frameworks and outperforms its supervised counterpart by a considerable margin without fine-tuning.
- Transformer-Based Models for Predicting Molecular Structures from Infrared Spectra Using Patch-Based Self-Attention29
A transformer-based model with a patch-based self-attention spectrum embedding layer, designed to prevent the loss of spectral information while maintaining simplicity and effectiveness, achieves state-of-the-art performance on simulated data sets and demonstrates proficiency in analyzing intricate and variable fingerprint regions.
- 28
- 27
- Ground-Aware Point Cloud Semantic Segmentation for Autonomous Driving25
The proposed ground-aware attention module captures long-range dependence between ground and objects, which significantly facilitates the segmentation of small objects that only consist of a few points in extremely sparse point clouds.
- 25
- 24
- 23
- 20
- 20
- LocalDyGS: Multi-view Global Dynamic Scene Modeling via Adaptive Local Implicit Feature Decoupling19
The method not only demonstrates competitive performance on various fine-scale datasets compared to state-of-the-art (SOTA) methods, but also represents the first attempt to model larger and more complex highly dynamic scenes.
- Facial Anatomical Landmark Detection Using Regularized Transfer Learning With Application to Fetal Alcohol Syndrome Recognition18
A new regularized transfer learning approach that exploits the knowledge of a network learned on large facial recognition datasets to improve model generalizability under limited training samples, and is advantageous to other approaches in the literature.
- Joint Image Denoising and Disparity Estimation via Stereo Structure PCA and Noise-Tolerant Cost18
A new joint framework is proposed that iteratively optimizes these two different tasks in a multiscale fashion and achieves superior performance than the state-of-the-art image denoising and disparity estimation methods.
- 18
- Multi-view Self-supervised Disentanglement for General Image Denoising17
By taking two different corrupted versions of the same image as input, the proposed Multi-view Self-supervised Disentanglement (MeD) approach learns to disentangle the latent clean features from the corruptions and recover the clean image consequently.
- Surface-SOS: Self-Supervised Object Segmentation via Neural Surface Representation16
Surface-SOS is the first self-supervised approach that leverages neural surface representation to break the dependence on large amounts of annotated data and strong constraints and surpasses supervised single-view baselines remarkably.
- 15
- Revisiting Implicit Neural Representations in Low-Level Vision13
This work revisits INR and investigates its application in low-level image restoration tasks including image denoising, super-resolution, inpainting, and deblurring, suggesting the superior performance of INR in several low- level vision tasks with limited resources, outperforming its counterparts by over 2dB.
- Differentiating Operator Skill During Routine Fetal Ultrasound Scanning Using Probe Motion Tracking13
A novel convolutional neural network-based deep learning framework is presented to model ultrasound probe motion in order to classify operator skill levels, that is invariant to operators’ personal scanning styles.
- CoMBO: Conflict Mitigation via Branched Optimization for Class Incremental Segmentation12
This work introduces a novel approach, Conflict Mitigation via Branched Optimization (CoMBO), designed to explicitly refine queries for new classes through lightweight, class-specific adapters, and develops two strategies to further mitigate the conflict following the branched structure.
- 11
- 9
- Tactile Sketch Saliency9
This paper proposes to jointly predict the tactile saliency, depth map and semantic category of a sketch in an end-to-end learning-based framework, and proposes to synthesize training data by leveraging a collection of 3D shapes with 3D tactile Saliency information.
- 9
- 7
- 7
- 7
- Edge Distraction-aware Salient Object Detection6
This work first formulate their edge features extraction process as a boundary-filling problem, and proposes to explore cross-scale holistic contextual connections between every position pair of high-level feature maps regardless of their distances across different scales, to produce distraction-free edge features.
- 6
- Adaptation Strategies for Applying AWGN-Based Denoiser to Realistic Noise6
This paper trained a deep network integrating noise estimating and denoiser with mixed Gaussian (AWGN) and Random Value Impulse Noise (RVIN) to adapt the model to realistic noises, and investigated multi-channel, multi-scale and super-resolution approaches.
- Cross-Task Representation Learning for Anatomical Landmark Detection5
The main idea of the proposed method is to retain the feature representations of the source model on the target task data, and to leverage them as an additional source of supervisory signals for regularizing the target model learning, thereby improving its performance under limited training samples.
- Anatomy-Aware Self-supervised Fetal MRI Synthesis from Unpaired Ultrasound Images5
This paper seeks to produce images with MRI-like appearance directly from clinical US images, based on an assumption that the US and MRI data share a similar anatomical latent space, and proposes an adversarial structural constraint to regularise the anatomical structures between the two modalities during the synthesis.
- DeHate : A Holistic Hateful Video Dataset for Explicit and Implicit Hate Detection4
This work presents DeHate, which is the largest hateful video dataset to date, and benchmark state-of-the-art models, including both uni-modal and multi-modal architectures, and identify persistent challenges in detecting subtle and context-dependent forms of hate.
- Disentangled Pre-training for Image Matting4
Extensive experimental evaluations show that the proposed approach outperforms both the state-of-the-art matting methods and other alternative self-supervised initialisation approaches by a large margin, and shows the robustness of the proposed approach over different backbone architectures.
- Multimodal Hate Detection Using Dual-Stream Graph Neural Networks3
This work proposes a novel multimodal dual-stream graph neural network model that employs a graph-based framework to systematically model structured relationships within and across modalities in hateful video classification and has strong explainability.
- Audio-visual modelling in a clinical setting3
Experimental analysis on a large-scale clinical multi-modal ultrasound video dataset show that the proposed novel representation learning method provides good transferable anatomical representations that boost the performance of automated downstream clinical tasks, even outperforming fully-supervised solutions.
- Inferring Attention Shifts for Salient Instance Ranking3
This paper proposes a new deep learning-based model to leverage both bottom-up and top-down attention mechanisms for saliency rank prediction, and achieves state-of-the-art performances on the salient object ranking task across multiple datasets.
- 3
- Coupled Rain Streak and Background Estimation via Separable Element-Wise Attention3
A coupled rain streak and background estimation network that explores the intrinsic relations between two tasks and outperforms state-of-the-arts on 5 existing synthesized rain datasets and the real-world scenarios, without extra multi-scale or recurrent structure.
- 3
- Reasoning-Aware Multimodal Fusion for Hateful Video Detection2
An innovative Reasoning-Aware Multimodal Fusion (RAMF) framework is proposed, which designs Local-Global Context Fusion (LGCF) to capture both local salient cues and global temporal structures, and proposes Semantic Cross Attention (SCA) to enable fine-grained multimodal semantic interaction.
- Dual Representation Learning From Fetal Ultrasound Video and Sonographer Audio2
This paper designs a network architecture capable of learning useful representations such as of anatomical features and structures while recognising the correlation between an US video scan and the sonographer’s speech, and introduces dual representation learning from US video and audio.
- 2
- A Kernel Density Estimation Based Quality Metric for Quality Assessment of Obstetric Ultrasound Video2
An automated quality assessment algorithm is proposed that applies an object detector to detect fetal anatomies within ultrasound videos that calculates a new quality metric that is compared to other metrics of quality such as the confidence of the object detector model.
- Mask-Guided Deformation Adaptive Network for Human Parsing2
The auxiliary task of human mask and edge detection is introduced to facilitate human parsing and can achieve superior performance against state-of-the-art methods on both single and multiple human parsing datasets.
- Quantised Transforming Auto-Encoders: Achieving Equivariance to Arbitrary Transformations in Deep Networks2
This work proposes an auto-encoder architecture whose embedding obeys an arbitrary set of equivariance relations simultaneously, such as translation, rotation, colour changes, and many others, and demonstrates results of successful re-rendering of transformed versions of input images on several synthetic and real datasets.
- 1
- 1
- Audio-Visual Separation with Hierarchical Fusion and Representation Alignment1
A representation alignment approach that aligns the latent features of the audio encoder with embeddings extracted from pre-trained audio models is proposed, showing that it reduces modality gap between the audio and visual modalities.
- 1
- 1
- 1
- –
- Dynamic in Static: Hybrid Visual Correspondence for Self-Supervised Video Object Segmentation–
HVC is presented, a hybrid static-dynamic visual correspondence framework for self-supervised VOS, that extracts pseudo-dynamic signals from static images, enabling an efficient and scalable VOS model.
- ReBac: Current-to-Past Residual Background Correction for Class-Incremental Semantic Segmentation–
A principled residual modeling framework, termed ReBac (Residual Background Correction), which explicitly models background evolution from a current-to-past perspective, and designs a synergistic objective combining a Pseudo-Background Binary Cross-Entropy Loss with Background Adaptation Losses, jointly supervising and stabilizing background refinement across steps.
- –
- What Can We Learn from Harry Potter? An Exploratory Study of Visual Representation Learning from Atypical Videos–
Focusing on three key tasks in open-world learning: out-of-distribution detection, novel category discovery, and zero-shot action recognition, it is found that even straightforward learning approaches with atypical data consistently improve performance across various settings.
- –
- –
- –
- Speed Co-Augmentation for Unsupervised Audio-Visual Pre-training–
Experimental results show that the proposed speed co-augmentation method significantly improves the learned representations when compared to vanilla audio-visual contrastive learning.
- Show from Tell: Audio-Visual Modelling in Clinical Settings–
Experimental evaluations on a large-scale clinical multi-modal ultrasound video dataset show that the proposed self-supervised method learns good transferable anatomical representations that boost the performance of automated downstream clinical tasks, even outperforming fully-super supervised solutions.
- Erratum to “Multi-View Face Synthesis via Progressive Face Flow”–
In the above article [1], unfortunately, Fig. 5 was not displayed correctly with many empty images, so the correct version is supplemented here.
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 2026-10-15. 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.