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
- Rosa H. M. ChanSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works67 from public data
- Involution: Inverting the Inherence of Convolution for Visual Recognition418
The proposed involution operator could be leveraged as fundamental bricks to build the new generation of neural networks for visual recognition, powering different deep learning models on several prevalent benchmarks, including ImageNet classification, COCO detection and segmentation, together with Cityscapes segmentation.
- Generative Adversarial Networks in Time Series: A Systematic Literature Review334
A classified GAN variants designed for time series related applications are proposed, in which GANs deal with discrete time series and continuous time series data.
- Generative Adversarial Networks in Computer Vision268
The objective is to provide an overview as well as a critical analysis of the status of GAN research in terms of relevant progress toward critical computer vision application requirements and discuss the most compelling applications in computer vision.
- 263
- ACTION-Net: Multipath Excitation for Action Recognition247
This work designs a generic and effective module that can be embedded into 2D CNNs to form a simple yet effective ACTION-Net with very limited extra computational cost.
- Avalanche: an End-to-End Library for Continual Learning225
The proposed Avalanche, an open-source end-to-end library for continual learning research based on PyTorch is designed to provide a shared and collaborative codebase for fast prototyping, training, and reproducible evaluation of continual learning algorithms.
- Are We Ready for Service Robots? The OpenLORIS-Scene Datasets for Lifelong SLAM199
The term lifelong SLAM is used here to address SLAM problems in an ever-changing environment over a long period of time and the OpenLORIS-Scene datasets are released, which are collected in real-world indoor scenes, for multiple times in each place to include scene changes in real life.
- MINE: Towards Continuous Depth MPI with NeRF for Novel View Synthesis194
The proposed MINE is a continuous depth generalization of the Multiplane Images (MPI) by introducing the NEural radiance fields (NeRF) to perform novel view synthesis and depth estimation via dense 3D reconstruction from a single image.
- Beyond Text-Visual Attention: Exploiting Visual Cues for Effective Token Pruning in VLMs171
This work proposes VisPruner, a plug-and-play method that utilizes visual cues for more effective token pruning in LVLMs, significantly outperforming existing methods based on text-visual attention.
- Generative Adversarial Networks: A Survey and Taxonomy.94
An in-depth review of GAN-related research in the literature is provided, and an account of the architecture-variant and loss-variants, which have been proposed to handle these three challenges from two perspectives are provided.
- OpenLORIS-Object: A Robotic Vision Dataset and Benchmark for Lifelong Deep Learning84
A new lifelong robotic vision dataset ("OpenLORIS-Object") collected via RGB-D cameras is provided and the results demonstrate that the object recognition task in the ever-changing difficulty environments is far from being solved and the bottlenecks are at the forward/backward transfer designs.
- Learning from Temporal Gradient for Semi-supervised Action Recognition79
This paper introduces temporal gradient as an additional modality for more attentive feature extraction in semi-supervised video action recognition and explicitly distills the fine-grained motion representations from temporal gradient and imposes consistency across different modalities.
- On Learning Contrastive Representations for Learning with Noisy Labels76
A novel contrastive regularization function is proposed to learn robust contrastive representations of data on which the classifier is hard to memorize the label noise under the CE loss and it is revealed that the learned representations keep information related to true labels and discard informationrelated to corrupted labels.
- MC-LLaVA: Multi-Concept Personalized Vision-Language Model58
Comprehensive experiments demonstrate that MC-LLaVA achieves impressive multi-concept personalized responses, paving the way for VLMs to become better user assistants.
- CVPR 2020 continual learning in computer vision competition: Approaches, results, current challenges and future directions57
The first Continual Learning in Computer Vision challenge held at CVPR in 2020 has been one of the first opportunities to evaluate different continual learning algorithms on a common hardware with a large set of shared evaluation metrics and 3 different settings based on the realistic CORe50 video benchmark.
- 47
- Learning the Superpixel in a Non-iterative and Lifelong Manner45
This work targets at redefining the CNN-based superpixel segmentation as a lifelong clustering task and pro-poses an unsupervised CNN- based method called LNS-Net, which can learn superpixel in a non-iterative and lifelong manner without any manual labels.
- Challenges in Task Incremental Learning for Assistive Robotics45
All the three most reported lifelong/continual learning algorithms have failed with the increase in encountered batches across various metrics with indistinguishable performance comparing to the naive training mode, highlighting the current challenges in lifelong object recognition for assistive robots to operate in real-world dynamic scene.
- Weakly Supervised Object Localization as Domain Adaption40
A novel perspective is provided that models WSOL as a domain adaption (DA) task, where the score estimator trained on the source/image domain is tested on the target/pixel domain to locate objects and a DA-WSOL pipeline is designed to better engage DA approaches into WSOL to enhance localization performance.
- 36
- Evaluating the Small-World-Ness of a Sampled Network: Functional Connectivity of Entorhinal-Hippocampal Circuitry36
An efficient framework is presented for reconstructing functional connectivity using experimental spike-train data and it is shown that it is feasible to adjust the estimated small-world-ness results based on the number of neurons recorded to provide a more accurate reference of the network property.
- Background-Aware Classification Activation Map for Weakly Supervised Object Localization32
This work proposes a novel mechanism called the background-aware classification activation map (B-CAM) to add background awareness for WSOL training, which trained a background classifier with image-level annotation to produce adaptive background scores when determining the binary localization mask.
- Unifying Nonlocal Blocks for Neural Networks28
By concerning the property of spectral, this paper proposes an efficient and robust spectral nonlocal block, which can be more robust and flexible to catch long-range dependencies when inserted into deep neural networks than the existing nonlocal blocks.
- Reduced-Rank Linear Dynamical Systems25
This work proposes Reduced-Rank Linear Dynamical Systems (RRLDS), to automatically retrieve the intrinsic dimensionality of the latent space during model learning and can robustly learn latent space from short-length, noisy, count-valued data and significantly improve the prediction performance over the state-of-the-art methods.
- 24
- CatNet: Class Incremental 3D ConvNets for Lifelong Egocentric Gesture Recognition20
This work demonstrates a lifelong 3D convolutional framework - c(C)la(a)ss increment(t)al net(Net)works (CatNet), which enables life-long learning for egocentric gesture video recognition by learning the feature representation of an exemplar set selected from previous class samples.
- TEAM-Net: Multi-modal Learning for Video Action Recognition with Partial Decoding17
This work introduces sampling the input for the network from partially decoded videos based on the GOP-level, and proposes a plug-and-play mulTi-modal lEArning Module (TEAM) for training the network using information from I-frames and P-frames in an end-to-end manner.
- CodeDance: A Dynamic Tool-integrated MLLM for Executable Visual Reasoning16
This work introduces CodeDance, which explores executable code as a general solver for visual reasoning, and empirically observe novel emergent behaviors during RL training, suggesting a general and scalable mechanism for executable visual reasoning.
- MT-ORL: Multi-Task Occlusion Relationship Learning14
A novel architecture called Occlusion-shared and Path-separated Network (OPNet) is proposed, which solves the first issue by exploiting rich occlusion cues in shared high- level features and structured spatial information in task-specific low-level features.
- Synthetic-Neuroscore: Using a neuro-AI interface for evaluating generative adversarial networks13
This work describes an evaluation metric, called Neuroscore, for evaluating the performance of GANs that more directly reflects psychoperceptual image quality through the utilization of brain signals and shows that including neural responses during the training phase of the network can significantly improve the prediction capability of the proposed model.
- ThinkGen: Generalized Thinking for Visual Generation12
ThinkGen is presented, the first think-driven visual generation framework that explicitly leverages MLLM's CoT reasoning in various generation scenarios, and proposes a separable GRPO-based training paradigm (SepGRPO), alternating reinforcement learning between the MLLM and DiT modules.
- IROS 2019 Lifelong Robotic Vision: Object Recognition Challenge [Competitions]12
This report summarizes the IEEE/ RSJ International Conference on Intelli gent Robots and Systems (IROS) 2019 Lifelong Robotic Vision Competition (Lifelong Object Recognition Chal lenge) with the data set, rules, methods, and results from the top eight finalists (of over 150 teams) (Figure 2).
- OpenLORIS-Object: A Dataset and Benchmark towards Lifelong Object Recognition12
A new lifelong robotic vision dataset (“OpenLORIS-Object”) collected via RGB-D cameras mounted on mobile robots is provided and a testbed of 9 state-of-the-art lifelong learning algorithms are provided for validating lifelong object recognition algorithms.
- Cross-frequency information transfer from EEG to EMG in grasping12
The authors' results demonstrate the capability of TE to capture the direct interaction between EEG and EMG, and reveal instantaneous decrease in information transfer from EEG to the high frequency component of EMG during the onset of movement.
- Video-KTR: Reinforcing Video Reasoning via Key Token Attribution10
Video-KTR is proposed, a modality-aware policy shaping framework that performs selective, token-level RL by combining three attribution signals: visual-aware tokens identified via counterfactual masking to reveal perceptual dependence, temporal-aware tokens detected through frame shuffling to expose temporal sensitivity, and high-entropy tokens signaling predictive uncertainty.
- 10
- Neuroscore: A Brain-inspired Evaluation Metric for Generative Adversarial Networks10
This work introduces an evaluation metric, called Neuroscore, for evaluating the performance of GANs that more directly reflects psychoperceptual image quality through the utilization of brain signals and shows that including neural responses during the training phase of the network can significantly improve the prediction capability of the proposed model.
- PDO-s3DCNNs: Partial Differential Operator Based Steerable 3D CNNs9
PDO-s3DCNNs are the most general steerable CNNs for 3D rotations, in the sense that they cover all common subgroups of SO (3) and their representations, while existing methods can only be applied to specific groups and representations.
- Inter-intra Variant Dual Representations for Self-supervised Video Recognition9
This paper finds that existing contrastive learning based solutions for self-supervised video recognition focus on inter-variance encoding but ignore the intra-Variance existing in clips within the same video, and proposes to learn dual representations for each clip to encode intra- Variance through a shuffle-rank pretext task.
- 8
- 8
- 8
- 6
- 6
- An Efficient and Flexible Spike Train Model Via Empirical Bayes5
Estimation of simulation data indicates that the new framework can accurately predict mean spike counts simulated from different models and recover the connectivity weights among neural populations and provide insight into both neural interactions and spiking behaviours of the neuronal populations.
- SteerVTE: Seamless Video Text Editing with Style and Glyph Control4
SteerVTE is introduced, a unified framework that substantially outperforms existing video editing baselines across text accuracy, style consistency, and temporal coherence and is proposed a glyph-aware spatial-focal loss and a three-stage progressive training curriculum that scales from image to video data.
- Newton design: designing CNNs with the family of Newton’s methods4
A unified framework for understanding and designing CNN architectures with the family of Newton’s methods, which is referred to as Newton design is proposed, and it is observed that the standard feedforward CNN model (PlainNet) solves an optimization problem via a kind of quasi-Newton method.
- 4
- Effective connectivity matrix for neural ensembles4
An efficient framework to study the directional interactions within the multiple-input multiple-output (MIMO) biological neural network from spiketrain data is presented and an Effective Connectivity Matrix (ECM) is developed to visualize excitatory and inhibitory connections within the neural network.
- ChainV: Atomic Visual Hints Make Multimodal Reasoning Shorter and Better3
ChainV is proposed, a framework that dynamically integrates visual hints into the reasoning process, thereby making multimodal reasoning shorter and better and introducing a consistency-based evaluation mechanism to assess the reliability of the chosen hint.
- 3
- Weighted Network Density Predicts Range of Latent Variable Model Accuracy3
A biologically plausible latent variable model was first fit to neural activity recorded via 2-photon microscopic calcium imaging in the murine primary visual cortex and results show LVM will achieve high accuracy in dense networks.
- EvoVid: Temporal-Centric Self-Evolution for Video Large Language Models2
This work proposes a temporal-centric self-evolving framework that enables Video-LLMs to improve directly from raw, unannotated videos, and introduces two complementary temporal-centric rewards: a temporal-aware Questioner reward that encourages temporally dependent question generation through temporal perturbation sensitivity, and a temporal-grounded Solver reward that provides automatic temporal supervision via inherent video segment localization.
- MammothModa: Multi-Modal Large Language Model2
This report introduces MammothModa, yet another multi-modal large language model (MLLM) designed to achieve state-of-the-art performance starting from an elementary baseline that consistently outperforms the state-of-the-art models, e.g., LLaVA-series, across main real-world visual language benchmarks without bells and whistles.
- NeMI: Unifying Neural Radiance Fields with Multiplane Images for Novel View Synthesis2
The NeMI is a general two-dimensional and imageconditioned extension of NeRF, and a continuous depth generalization of MPI, which combines Neural radiance fields with Multiplane Images to perform novel view synthesis and depth estimation via dense 3D reconstruction from a single image.
- 2
- 2
- TextSculptor: Training and Benchmarking Scene Text Editing1
This work develops an automated data construction pipeline that combines text-aware image synthesis with programmatic text rendering and compositing and designs a tailored protocol that measures text accuracy, visual quality, and background preservation through OCR-based text alignment, multimodal judgment, and background-region similarity.
- A Neuro-AI Interface: Learning DNNs from the Human Brain1
The concept of a neuro-AI interface is introduced, which aims to use human’s neural responses as supervised information for helping AI systems solve a task that is difficult when using traditional machine learning strategies.
- 1
- –
- –
- AD-MIR: Bridging the Gap from Perception to Persuasion in Advertising Video Understanding via Structured Reasoning–
AD-MIR, a framework designed to decode advertising intent via a two-stage architecture, achieves state-of-the-art performance, surpassing the strongest general-purpose agent, DVD, by 1.8% in strict and 9.5% in relaxed accuracy.
- –
- –
- –
- A modified hierarchical graph cut based video segmentation approach for high frame rate video–
The novelty of the research presented here is to utilize a fast object tracking algorithm conjoined with graph cut based segmentation in a hierarchical framework to concentrate on the high variations of the video data and use modified hierarchical processing to capture the spatiotemporal variation.
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 2026-10-12. 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.