Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks74 from public data
- 259
- 153
- Underwater Species Detection using Channel Sharpening Attention93
A real-world underwater object detection dataset (UODD), which covers more than 3K images of the most common aquatic products, and Channel Sharpening Attention Module (CSAM) as a plug-and-play module to further fuse high-level image information, providing the network with the privilege of selecting feature maps.
- Cross-SRN: Structure-Preserving Super-Resolution Network With Cross Convolution83
A hierarchical feature exploitation network to probe and preserve structural information in a multi-scale feature fusion manner, invoking a lightweight structure-preserving network named as Cross-SRN, which outperforms the state-of-the-art methods on the selected benchmark.
- Novel Coplanar Line-Points Invariants for Robust Line Matching Across Views46
A new coplanar line-points invariant upon a newly developed projective invariant, named characteristic number, is constructed and a line matching algorithm using the invariant is proposed, rendering more robust matching across views.
- Learning Pixel-wise Alignment for Unsupervised Image Stitching37
An unsupervised image stitching framework, breaking through the coplanar constraints in homography estimation, yielding accurate pixel-wise alignment under limited overlapping regions, achieves state-of-the-art performance that surpasses both traditional and deep learning approaches by a large margin.
- Hierarchical projective invariant contexts for shape recognition37
A new shape descriptor (HCNC) invariant to projective deformations, compact with a hierarchical strategy, rendering efficient matching and robust to noise, part-missing and articulated deformations.
- 29
- Line matching based on line-points invariant and local homography27
A novel line-points invariant based on a new projective invariant named characteristic number is presented, able to reflect the intrinsic geometry of line and points, which keeps invariant across views.
- Region-based CNN for Logo Detection27
This paper aims to experimentally investigate the appropriate architecture and settings of R-CNN for logo detection, by comparing several popular frameworks ofR-CNN and considering the characteristics of logo objects in images.
- 3D facial landmark localization using texture regression via conformal mapping26
A novel 3D facial landmark localization algorithm that can convert a 3D model to 2D using both geometry and texture information is proposed, which is impervious to pose changes and robust with respect to changes in expression.
- Unseen No More: Unlocking the Potential of CLIP for Generative Zero-shot HOI Detection24
This work presents the first generation-based model using CLIP for zero-shot HOI detection, coined HOIGen, and develops a CLIP-injected feature generator in accordance with the generation of human, object and union features that allows to unlock the potential of CLIP for feature generation instead of feature extraction only.
- Multi-Layer Sparse Representation for Weighted LBP-Patches Based Facial Expression Recognition23
A multi-layer sparse representation framework is proposed for multi-intensity facial expression recognition, especially for low-intensity expressions and noisy expressions in reality, which is a critical problem but seldom addressed in the existing works.
- 15
- Hierarchical Similarity Learning for Aliasing Suppression Image Super-Resolution15
The self-similarity is considered and a hierarchical image super-resolution network (HSRNet) is proposed to suppress the influence of aliasing and the experimental result shows that HSRNet achieves better quantitative and visual performance than other works and remits the aliasing more effectively.
- Coupling Deep Textural and Shape Features for Sketch Recognition15
This paper explicitly explores the shape properties of sketches, which has almost been neglected before in the context of deep learning, and proposes a sequential dual learning strategy that combines both shape and texture features.
- 11
- 11
- 9
- 9
- 9
- A rotation robust shape transformer for cartoon character recognition8
A rotation robust shape transformer for cartoon character recognition, which beats both handcrafted and deep learning methods on the proposed challenging Cartoon dataset and traditional datasets, and develops an online cartoon character recognition application for animation scenarios.
- 8
- FARS: A Fully Automated Research System Deployed at Scale7
FARS (Fully Automated Research System), a fully automated AI-for-AI research system designed to operate across research topics at scale, is presented and reviews indicate that FARS can produce review-worthy and occasionally strong AI/ML research artifacts in a large-scale public deployment.
- 7
- Two Teachers Are Better Than One: Semi-supervised Elliptical Object Detection by Dual-Teacher Collaborative Guidance6
A new semi-supervised teacher-student framework, namely Dual-Teacher Collaborative Guidance (DTCG), comprising a five-parameter teacher detector, a six-parameter teacher detector, and a student detector that allows the two teachers specializing in different regression approaches, and co-instructing the student within a unified model, so as to prevent errors and enhance performance.
- 6
- Characteristic Number: Theory and Its Application to Shape Analysis6
A projective invariant named after the characteristic number of planar algebraic curves, which reveals an intrinsic property of an algebraic hypersurface or curve, which relies no more on the existence of the surface or curve as its planar version.
- 6
- 5
- 5
- 5
- An Intelligent 3D Printing Path Planning Algorithm 3D Printing Path Planning Algorithm5
The sub-partition-based intelligent path planning method provides a new idea for the additive manufacturing process of large industrial parts by significantly reduced the number of paths, the idle travel distance, and thenumber of print head lifts.
- A new geometric descriptor for symbols with affine deformations5
A new affine invariant, named characteristic ratio (CHAR) that includes more points for complex symbols with rich inner structures, and a novel shape descriptor with CHAR values calculated on collinear points that cover the convex hull of a shape are proposed.
- 5
- Q-Bench-Portrait: Benchmarking Multimodal Large Language Models on Portrait Image Quality Perception4
Q-Bench-Portrait is introduced, the first holistic benchmark specifically designed for portrait image quality perception, comprising 2,765 image-question-answer triplets and revealing that although current models demonstrate some competence in portrait image perception, their performance remains limited and imprecise, with a clear gap relative to human judgments.
- EvolMem: A Cognitive-Driven Benchmark for Multi-Session Dialogue Memory4
EvolMem is a new benchmark for assessing multi-session memory capabilities of LLMs and agent systems that is grounded in cognitive psychology and encompasses both declarative and non-declarative memory, further decomposed into multiple fine-grained abilities.
- 4
- Bag of Spatial Visual Words Model for Scene Classification4
An approach to recognize scene categories by means of a novel model named bag of spatial visual words was proposed, which obtained a classification result using LIBSVM and two kinds of features were used in the experiments:"V1-like" filters and PACT features.
- 3
- MIPI 2023 Challenge on RGBW Fusion: Methods and Results3
This paper summarizes and reviews the RGBW Joint Fusion and Denoise track on MIPI 2023 and describes the models developed in this challenge, including PSNR, SSIM, LPIPS, and KLD.
- Online gesture-based interaction with visual oriental characters based on manifold learning3
A new feature descriptor named Segmented Directed-edge Vector (SDV) is proposed, able to capture the characteristics of visual oriental characters, and is explicitly build the mappings from SDVs to features in a subspace by a modified Locality Preserving Projections (LPP) method with stroke class constraints.
- UniDial-EvalKit: A Unified Toolkit for Evaluating Multi-Faceted Conversational Abilities2
This work presents UniDial-EvalKit (UDE), a unified evaluation toolkit for assessing interactive AI systems that standardizes heterogeneous data formats into a universal schema, streamlines complex evaluation pipelines through a modular architecture, and aligns metric calculations under a hierarchical scoring aggregation.
- 2
- LatentRevise: Learning from Zero-Hit Reasoning2
LatentRevise is introduced, a first-order latent revision method that recovers training signal for this zero-hit regime of reinforcement learning with verifiable rewards and finds that continuations from the revised prefix lengthen, exhibit self-reflection, and reach correct answers missed by the original rollouts.
- Droplet3D: Commonsense Priors from Videos Facilitate 3D Generation2
This paper introduces Droplet3D-4M, the first large-scale video dataset with multi-view level annotations, and introduces Droplet3D, a generative model supporting both image and dense text input, which exhibits the potential for extension to scene-level applications.
- 2
- Semi-supervised Source Detection in Astronomical Images: New Benchmark and Strong Baseline1
A novel semi-supervised learning framework coined Nova Teacher, capable of detecting dense sources effectively given sparse annotations, and competes against other detectors on a natural image dataset, validating its generalization ability to various scenarios.
- 1
- As Pseudo-Label Free as Possible: Leveraging Adaptive Feature Generation for Sparsely Annotated Object Detection1
A novel Adaptive Feature Generation (AdaptFG) model that generates features based on class names that consistently outperforms state-of-the-art SAOD methods on the PASCAL VOC and MS COCO benchmarks.
- 1
- 1
- 1
- 1
- 1
- 1
- Cross-view action matching using a novel projective invariant on non-coplanar space-time points1
A new projective invariant named characteristic number (CN) is used, which can be used to describe 3D non-coplanar points, and promising recognition results indicate the invariance to varying viewpoints compared with the state-of-the-art.
- 1
- 1
- Time Series Class-Incremental Learning via Confidence-guided Mask Distillation and Prototype-guided Contrastive Learning–
This work proposes a novel confidence-guided mask distillation (CMD) framework, and develops prototype-guided contrastive learning (PCL) to alleviate the classifier bias on new classes, through extra contrastive constraints to push away the feature distributions of old feature prototypes from those of new classes features.
- –
- –
- Weakly supervised artificial intelligence for multi-cancer detection of lymph node metastasis on whole slide images–
It is shown that MambaMIL+HiLA-MIL is a multi-cancer four-class lymph node metastasis model, demonstrating high efficiency and robustness across multiple centers and various feature extractors, offering a reliable tool for clinical lymph node metastasis classification.
- –
- –
- –
- –
- –
- –
- Notice of Removal An improved GPGPU-Accelerated parallelization for rotation invariant thinning algorithm–
This paper presents a meta-modelling system that automates the very labor-intensive and therefore time-heavy and therefore expensive and expensive process of manually cataloging and cataloging books and manuscripts for publication and distribution.
- Distributed Compressed Sensing MRI Using Volume Array Coil–
A singular value decomposition (SVD) based sparsity basis was developed for the CS-MRI with a volume array coil configuration, and the experimental results showed that the target images could still be faithfully reconstructed using the proposed method, which offered a better imaging performance compared to conventional CS schemes.
- Image categorization based on spatial visual vocabulary model–
This paper hierarchically divide images into sub regions and construct the spatial visual vocabulary by grouping the low-level features collected from every corresponding spatial sub region into a specified number of clusters using k-means algorithm.
- –
- –
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.