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
- Qiong LiuSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works154 from public data
- MobileFaceNets: Efficient CNNs for Accurate Real-Time Face Verification on Mobile Devices771
A class of extremely efficient CNN models, MobileFaceNets, which use less than 1 million parameters and are specifically tailored for high-accuracy real-time face verification on mobile and embedded devices, achieve significantly improved efficiency over previous state-of-the-art mobile CNNs.
- TimeTraveler: Reinforcement Learning for Temporal Knowledge Graph Forecasting242
This work proposes the first reinforcement learning method for forecasting that defines a relative time encoding function to capture the timespan information, and proposes a novel representation method for unseen entities to improve the inductive inference ability of the model.
- 217
- Learning Neural Ordinary Equations for Forecasting Future Links on Temporal Knowledge Graphs184
A novel continuum model is proposed by extending the idea of neural ordinary differential equations to multi-relational graph convolutional networks and encodes both temporal and structural information into continuous-time dynamic embeddings, showing its superior performance on the future link forecasting task.
- 173
- 160
- 115
- DyERNIE: Dynamic Evolution of Riemannian Manifold Embeddings for Temporal Knowledge Graph Completion104
Dy- ERNIE is proposed, a non-Euclidean embedding approach that learns evolving entity representations in a product of Riemannian manifolds, where the composed spaces are estimated from the sectional curvatures of underlying data.
- 98
- Ultra-dense GAN for satellite imagery super-resolution89
An ultra-dense GAN (udGAN) is proposed for image SR, where the internal layout of the residual block is reformed into a two-dimensional matrix topology, which can provide additional diagonal connections so that the model can still accomplish enough pathways with fewer layers.
- 76
- 73
- 67
- 65
- Unsupervised Image Super-Resolution with an Indirect Supervised Path57
A novel framework which is composed of two stages: unsupervised image translation between real LR and synthetic LR images; and supervised super-resolution from approximated real LR images to the paired HR images, which achieves very good performance on datasets of NTIRE 2017, NTIRE 2018 and NTIRE 2020.
- SCEdit: Efficient and Controllable Image Diffusion Generation via Skip Connection Editing47
An efficient generative tuning framework, dubbed SCEdit, which integrates and edits Skip Connection using a lightweight tuning module named SC-Tuner, and allows for straightforward extension to controllable image syn-thesis by injecting different conditions with Controllable SC-Tuner.
- 47
- When Face Recognition Meets Occlusion: A New Benchmark44
A simulated occlusion face recognition dataset that combines original normal face images and occluded face images to form the final dataset, termed as Webface-OCC, which significantly outperforms the state-of-the-arts face recognition models trained on existing datasets.
- 41
- Time-dependent Entity Embedding is not All You Need: A Re-evaluation of Temporal Knowledge Graph Completion Models under a Unified Framework38
This work provides the first unified open-source framework for temporal KG completion models with full composability, where temporal embeddings, score functions, loss functions, regularizers, and the explicit modeling of reciprocal relations can be combined arbitrarily.
- Video Similarity and Alignment Learning on Partial Video Copy Detection37
The Video Similarity and Alignment Learning (VSAL) approach, which jointly models spatial similarity, temporal similarity and partial alignment, achieves the state-of-the-art F1-score on VCDB core dataset and constructs a new benchmark of partial video copy detection and localization.
- 37
- A Simple But Powerful Graph Encoder for Temporal Knowledge Graph Completion35
This paper proposes a simple but powerful graph encoder for TKGC, called TARGCN, which extensively explores every entity's temporal context for learning contextualized representations and finds that instead of adopting various kinds of complex modules, it is more beneficial to efficiently capture the temporal contexts of entities.
- Can Multimodal Large Language Models Truly Perform Multimodal In-Context Learning?33
A simple yet effective approach is proposed, termed Mixed Modality In-Context Example Selection (MMICES), which considers both visual and language modalities when selecting demos, which revealed that multimodal ICL is predominantly driven by the textual content whereas the visual information in the demos has little influence.
- 30
- 29
- 26
- Face image super-resolution through locality-induced support regression26
A novel face image super-resolution method named Locality-induced Support Regression (LiSR), which defines the support in the LR or HR patch manifold space by incorporating the locality constraint, which can well preserve the manifold structure of the training set.
- 25
- Efficient single image super-resolution via graph-constrained least squares regression25
The gCLSR, namely graph-Constrained Least Squares Regression, to super-resolve a high- resolution (HR) image from a single low-resolution (LR) observation while preserving the intrinsic geometric structure of the original HR image patch manifold.
- 23
- 22
- Ship Instance Segmentation Based on Rotated Bounding Boxes for SAR Images22
The proposed SRNet achieved more accurate ship instance masks in SAR images and a new adaptive Intersection-over-Union threshold (AIoU) was proposed based on the aspect-ratio information of ships to raise positive samples to improve the effectiveness of training.
- 20
- 20
- 19
- 18
- 18
- 18
- 17
- Semantic boundary enhancement and position attention network with long-range dependency for semantic segmentation16
A semantic boundary enhancement and position network (SBEPNet) is proposed that enhances feature maps with semantic boundaries and captures useful contextual information along these boundaries.
- 16
- Multi-Stage Degradation Homogenization for Super-Resolution of Face Images With Extreme Degradations15
- A novel face super resolution approach for noisy images using contour feature and standard deviation prior15
Different from the existing methods which directly enhance the noisy input image in intensity feature space, the proposed method introduces a contour feature which is robust to noise and which outperforms most classic methods both quantitatively and qualitatively.
- Improved fast intra prediction algorithm of H.264/AVC14
An improved fast intra prediction algorithm including block type selection and mode decision algorithm based on analysis of edge feature of a block and filters out unlikely block type and candidate modes to reduce the RDO calculations is proposed.
- 13
- 13
- 13
- Noise Face Image Hallucination via Data-Driven Local Eigentransformation11
Experimental results show the effectiveness of the proposed Data-driven Local Eigentransformation DLE method for face hallucination by exploiting the local geometry structure of data manifold and learning a specified eigentransformed model for each observed image.
- Coupled Discriminant Multi-Manifold Analysis with Application to Low-Resolution Face Recognition11
Experimental results conducted on two standard face recognition databases demonstrate the superiority of the proposed coupled discriminant multi-manifold analysis (CDMMA).
- Noise robust face hallucination employing Gaussian–Laplacian mixture model11
A moderately sparse prior model referred to as a Gaussian-Laplacian mixture (GLM) distribution is suggested and employed to infer the optimal solution under the Bayesian framework and not only maintains same hallucination performance as SR under noise free scenarios but also outperforms the latter remarkably in the presence of noise.
- Single image de-raining via clique recursive feedback mechanism10
A new rain removal network based on the clique recursive feedback mechanism is proposed, considering the interaction of the feature between different convolution layers, and constructing a residual clique block (RCB) to infer the local information, allowing the network to rectify the model parameters in RCB alternately.
- Face super resolution based on parent patch prior for VLQ scenarios10
The target patch as well as the surrounding pixels, which is called parent patch, is used to represent the target patch and by incorporating the peripheral information, the parent patch is much more robust to noise in the LR and HR consistency learning.
- 10
- 10
- Face hallucination with shape parameters projection constraint10
A face hallucination with shape parameters projection constraint is proposed, which outperforms the traditional schemes significantly both in subjective and objective quality.
- 9
- Local Eyebrow Feature Attention Network for Masked Face Recognition9
A feature fusion model based on graph convolutional network (GCN) is proposed to learn the feature association structure of eye features, brow features, and global facial features to improve the recognition performance of existing face recognition technology on masked faces.
- Two-step superresolution approach for surveillance face image through radial basis function-partial least squares regression and locality-induced sparse representation9
A comparison of some state-of-the-art SR methods shows the superiority of the proposed two-step approach, RBF-PLS global face regression followed by LiSR-based local patch reconstruction, which demonstrates the effectiveness under both simulation conditions and some real conditions.
- 8
- 8
- Heteroskedasticity tuned mixed-norm sparse regularization for face hallucination8
This paper derives the locally weighted variants of standard regularization representation from Bayesian inference perspective, which impose the similarity constraint within the observed image and training images onto the penalty function to deal with noise robust face hallucination.
- Face Image Superresolution via Locality Preserving Projection and Sparse Coding8
A novel sparse representation based face image superresolution (SR) method is proposed to reconstruct a high resolution (HR) face image from a LR observation by integrating the hallucinated HR patches together with overlapped in adjacent to obtain the final HR face image.
- Low-Resolution and Low-Quality Face Super-Resolution in Monitoring Scene via Support-Driven Sparse Coding8
A novel sparse coding method via exploiting the support information on the coding coefficients by solving a weighted sparse problem based on the distances between the input patch and bases in the dictionary is presented.
- 7
- 7
- 7
- 6
- 6
- Face Image Super-Resolution Through Improved Neighbor Embedding6
A novel face image super-resolution method based on Tikhonov Regularized Neighbor Representation, which is called TRNR for short is proposed, which plays a better subjective and objective performance, which can recover more fine structures and details from an input low-resolution image, when compared to previously reported methods.
- 6
- 6
- 6
- 6
- 6
- 5
- 5
- 5
- HRM graph constrained dictionary learning for face image super-resolution5
A novel sparse coding based face image SR method to reconstruct a HR face image from a Low Resolution (LR) observation, which uses graph constructed in HR Manifold (HRM) as regularization term and achieves better quality when compared with other state of theart methods.
- 5
- Face image super-resolution via nearest feature line5
A manifold learning based algorithm using 'Nearest Feature Line - NFL' to hallucinate high-resolution face image and select neighborhood samples for each input LR image in order to enhance the representation capacity of available face samples and reduce the computational complexity.
- 4
- 4
- 4
- 4
- Nearest feature line embedding for face hallucination4
Using the NFL metric to define the neighbourhood relations between face samples to improve the expressing power of the given training samples for reconstruction and Experimental results demonstrate that the method is effective at preserving detailed visual information.
- 4
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 1
- IDRetracor: Towards Visual Forensics against Malicious Face Swapping1
This work proposes IDRetracor, a model designed to recover arbitrary original target identities from fake faces generated by various face swapping techniques, and introduces Mapping-Aware Convolutions (MACs), which consist of multiple dynamically combined kernels guided by the mapping resolver to adaptively handle diverse face swapping patterns.
- Cross-Modal Face Super-Resolution Based on Quasi-Siamese Domain Transfer Fusion Network1
A Quasi-Siamese Domain Transfer Fusion Network (QSDTFN) for CMFSR is proposed, whose two branches transfer two LR face modality to HR face modality by domain transfer, respectively.
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- Face Hallucination Based On Sample Selection Bias Correction1
Experiments show that the face super-resolution algorithm based on sample selection bias correction can improve the subjective and objective quality of the input low resolution face images compared to traditional eigenface algorithm.
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Video-Based Person Re-Identification With Unregulated Sequences–
An adaptive unregulated sub sequence detection and refinement method is proposed to purify original video sequence and obtain a more effective and discriminative feature representation for video-based person re-id.
- –
- –
- –
- –
- –
- –
- –
- –
- –
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.