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
- Zhuowen TuSuggested from co-authorship
- Longin Jan LateckiSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works400 from public data
- DOTA: A Large-Scale Dataset for Object Detection in Aerial Images3,264
A large-scale Dataset for Object deTection in Aerial images (DOTA) is introduced and state-of-the-art object detection algorithms on DOTA are evaluated, demonstrating that DOTA well represents real Earth Vision applications and are quite challenging.
- An End-to-End Trainable Neural Network for Image-Based Sequence Recognition and Its Application to Scene Text Recognition3,166
A novel neural network architecture, which integrates feature extraction, sequence modeling and transcription into a unified framework, is proposed, which generates an effective yet much smaller model, which is more practical for real-world application scenarios.
- Gliding Vertex on the Horizontal Bounding Box for Multi-Oriented Object Detection1,012
An obliquity factor based on area ratio between the object and its horizontal bounding box, guiding the selection of horizontal or oriented detection for each object is introduced, and five extra target variables are added to the regression head of faster R-CNN, which requires ignorable extra computation time.
- TextBoxes: A Fast Text Detector with a Single Deep Neural Network955
An end-to-end trainable fast scene text detector, named TextBoxes, which detects scene text with both high accuracy and efficiency in a single network forward pass, involving no post-process except for a standard non-maximum suppression.
- 905
- TextBoxes++: A Single-Shot Oriented Scene Text Detector897
An end- to-end trainable fast scene text detector, named TextBoxes++, which detects arbitrary-oriented scene text with both high accuracy and efficiency in a single network forward pass, and significantly outperforms the state-of-the-art approaches for word spotting and end-to-end text recognition tasks on popular benchmarks.
- Real-Time Scene Text Detection with Differentiable Binarization891
This paper proposes a module named Differentiable Binarization (DB), which can perform the binarization process in a segmentation network, and validate the performance improvements of DB on five benchmark datasets, which consistently achieves state-of-the-art results, in terms of both detection accuracy and speed.
- Detecting Oriented Text in Natural Images by Linking Segments775
SegLink, an oriented text detection method to decompose text into two locally detectable elements, namely segments and links, achieves an f-measure of 75.0% on the standard ICDAR 2015 Incidental (Challenge 4) benchmark, outperforming the previous best by a large margin.
- Asymmetric Non-Local Neural Networks for Semantic Segmentation720
Asymmetric Pyramid Non- local Block (APNB) and Asymmetric Fusion Non-local Block (AFNB) are presented, which has two prominent components: APNB leverages a pyramid sampling module into the non-local block to largely reduce the computation and memory consumption without sacrificing the performance.
- Object Detection in Aerial Images: A Large-Scale Benchmark and Challenges714
The expanded large-scale DOTA dataset, the extensive baselines, the code library and the challenges can facilitate the designs of robust algorithms and reproducible research on the problem of object detection in aerial images.
- End-to-End Semi-Supervised Object Detection with Soft Teacher698
This paper presents an end-to-end semi-supervised object detection approach, in contrast to previous more complex multi-stage methods, and proposes two simple yet effective techniques within this framework: a soft teacher mechanism where the classification loss of each unlabeled bounding box is weighed by the classification score produced by the teacher network.
- Robust Scene Text Recognition with Automatic Rectification693
RARE (Robust text recognizer with Automatic REctification), a recognition model that is robust to irregular text, which is end-to-end trainable, requiring only images and associated text labels, making it convenient to train and deploy the model in practical systems.
- Mask TextSpotter: An End-to-End Trainable Neural Network for Spotting Text with Arbitrary Shapes650
The recognition module of the Mask TextSpotter method is investigated separately, which significantly outperforms state-of-the-art methods on both regular and irregular text datasets for scene text recognition.
- Multi-oriented Text Detection with Fully Convolutional Networks596
A novel approach for text detection in natural images that consistently achieves the state-of-the-art performance on three text detection benchmarks: MSRA-TD500, I CDAR2015 and ICDAR2013.
- OCRBench: on the hidden mystery of OCR in large multimodal models574
A comprehensive evaluation of large multimodal models, such as GPT4V and Gemini, in various text-related visual tasks, revealing both the strengths and weaknesses of these models, particularly in handling multilingual text, handwritten text, non-semantic text, and mathematical expression recognition.
- Rotation-Sensitive Regression for Oriented Scene Text Detection561
The proposed method named Rotation-sensitive Regression Detector (RRD) achieves state-of-the-art performance on several oriented scene text benchmark datasets, including ICDAR 2015, MSRA-TD500, RCTW-17, and COCO-Text, and achieves a significant improvement on a ship collection dataset, demonstrating its generality on oriented object detection.
- Revisiting multiple instance neural networks533
This article revisit Multiple Instance Neural Networks (MINNs) that the neural networks aim at solving the MIL problems and proposes a new type of MINN to learn bag representations, which is different from the existing MINNs that focus on estimating instance label.
- EPNet: Enhancing Point Features with Image Semantics for 3D Object Detection514
A novel fusion module is proposed to enhance the point features with semantic image features in a point-wise manner without any image annotations to address two critical issues in the 3D detection task, including the exploitation of multiple sensors~ and the inconsistency between the localization and classification confidence.
- 496
- YOLOP: You Only Look Once for Panoptic Driving Perception495
To the best knowledge, this is the first work that can process these three visual perception tasks simultaneously in real-time on an embedded device Jetson TX2(23 FPS), and maintain excellent accuracy.
- Side Adapter Network for Open-Vocabulary Semantic Segmentation487
This paper presents a new framework for open-vocabulary semantic segmentation with the pre-trained vision-language model, named Side Adapter Network (SAN), which significantly outperforms other counterparts, with up to 18 times fewer trainable parameters and 19 times faster inference speed.
- ICDAR2019 Competition on Scanned Receipt OCR and Information Extraction452
The ICDAR 2019 Challenge on "Scanned receipts OCR and key information extraction" (SROIE) covers important aspects related to the automated analysis of scanned receipts, and is considered to evolve into a useful resource for the community, drawing further attention and promoting research and development efforts in this field.
- Real-Time Scene Text Detection With Differentiable Binarization and Adaptive Scale Fusion431
A Differentiable Binarization (DB) module that integrates the binarization process, one of the most important steps in the post-processing procedure, into a segmentation network is proposed and an efficient Adaptive Scale Fusion (ASF) module is proposed to improve the scale robustness by fusing features of different scales adaptively.
- TANet: Robust 3D Object Detection from Point Clouds with Triple Attention428
A novel TANet is introduced in this paper, which mainly contains a Triple Attention (TA) module, and a Coarse-to-Fine Regression (CFR) module that boosts the accuracy of localization without excessive computation cost.
- 425
- 410
- Scene text detection and recognition: recent advances and future trends402
This literature review can serve as a good reference for researchers in the areas of scene text detection and recognition and identify state-of-the-art algorithms, and predict potential research directions in the future.
- Triplet-Center Loss for Multi-view 3D Object Retrieval389
This paper study variants of deep metric learning losses for 3D object retrieval, which did not receive enough attention from this area, and proposes a novel loss named triplet-center loss, which can further enhance the discriminative power of the features.
- Multi-oriented Scene Text Detection via Corner Localization and Region Segmentation389
This paper proposes to detect scene text by localizing corner points of text bounding boxes and segmenting text regions in relative positions and achieves better or comparable results in both accuracy and efficiency.
- 377
- Progressive Pose Attention Transfer for Person Image Generation375
A new generative adversarial network to the problem of pose transfer, i.e., transferring the pose of a given person to a target one, which can generate training images for person re-identification, alleviating data insufficiency.
- A Simple Baseline for Open-Vocabulary Semantic Segmentation with Pre-trained Vision-Language Model360
This paper refuses the use of the prevalent one-stage FCN based framework, and advocates a two-stage semantic segmentation framework, with the first stage extracting generalizable mask proposals and the second stage leveraging an image based CLIP model to perform open-vocabulary classification on the masked image crops which are generated in the firststage.
- TextField: Learning a Deep Direction Field for Irregular Scene Text Detection345
A novel text detector named TextField, which outperforms the state-of-the-art methods by a large margin on two curved text datasets: Total-Text and SCUT-CTW1500, respectively; TextField also achieves very competitive performance on multi-oriented datasets: ICDAR 2015 and MSRA-TD500.
- 343
- Scalable Person Re-identification on Supervised Smoothed Manifold342
An unconventional manifold-preserving algorithm is proposed, which can make best use of supervision from training data, whose label information is given as pairwise constraints, scale up to large repositories with low on-line time complexity, and be plunged into most existing algorithms, serving as a generic postprocessing procedure to further boost the identification accuracies.
- End-to-End Temporal Action Detection With Transformer341
TadTR is an end-to-end Transformer-based method for temporal action detection that achieves state-of-the-art performance on THUMOS14 and HACS Segments, and requires lower computation cost than previous detectors, while preserving remarkable performance.
- 310
- GIFT: A Real-Time and Scalable 3D Shape Search Engine298
The proposed 3D shape search engine, which combines GPU acceleration and Inverted File (Twice), is named as GIFT, which outperforms the state-of-the-art methods significantly in retrieval accuracy on various shape benchmarks and competitions.
- 278
- TransCrowd: weakly-supervised crowd counting with transformers268
This paper proposes TransCrowd, which reformulates the weakly-supervised crowd counting problem from the perspective of sequence-to-count based on transformers, and is the first work to adopt a pure transformer for crowd counting research.
- Deep-Person: Learning discriminative deep features for person Re-Identification265
This paper proposes to apply Long Short-Term Memory in an end-to-end way to model the pedestrian, seen as a sequence of body parts from head to foot, and develops a novel three-branch framework named Deep-Person, which learns highly discriminative features for person Re-ID.
- ICDAR2017 Competition on Reading Chinese Text in the Wild (RCTW-17)256
This report introduces RCTW, a new competition that focuses on Chinese text reading with a large-scale dataset with over 12,000 annotated images and calls for more future research on the Chinese textReading problem.
- Scene Text Recognition from Two-Dimensional Perspective251
A simple yet effective model, called Character Attention Fully Convolutional Network (CA-FCN), is devised for recognizing the text of arbitrary shapes with a semantic segmentation network, where an attention mechanism for characters is adopted.
- Traffic sign detection and recognition using fully convolutional network guided proposals241
A novel framework with two deep learning components including fully convolutional network (FCN) guided traffic sign proposals and deep Convolutional neural network (CNN) for object classification to perform fast and accurate traffic sign detection and recognition.
- Non-stationary texture synthesis by adversarial expansion230
This paper proposes a new approach for example-based non-stationary texture synthesis that uses a generative adversarial network (GAN), trained to double the spatial extent of texture blocks extracted from a specific texture exemplar, and demonstrates that it can cope with challenging textures, which no other existing method can handle.
- Occluded Video Instance Segmentation: A Benchmark229
A simple plug-and-play module that performs temporal feature calibration to complement missing object cues caused by occlusion is presented, and a remarkable AP improvement on the OVIS dataset is obtained.
- 220
- Mask TextSpotter v3: Segmentation Proposal Network for Robust Scene Text Spotting216
The proposed Mask TextSpotter v3 is an end-to-end trainable scene text spotter that adopts a Segmentation Proposal Network (SPN) that is anchor-free and gives accurate representations of arbitrary-shape proposals, superior to RPN in detecting text instances of extreme aspect ratios or irregular shapes.
- 214
- An End-to-End Transformer Model for Crowd Localization213
This paper proposes an elegant, end-to-end Crowd Localization Transformer named CLTR that solves the task in the regression-based paradigm, and introduces a KMO-based Hungarian matcher, which adopts the nearby context as the auxiliary matching cost.
- ICDAR 2019 Robust Reading Challenge on Reading Chinese Text on Signboard209
A large-scale dataset of 25,000 annotated signboard images, in which all the text lines and characters are annotated with locations and transcriptions, were released and a multi ground truth (multi-GT) evaluation method was proposed to make evaluation fairer.
- MASTER: Multi-aspect non-local network for scene text recognition196
The MASTER is proposed, a self-attention based scene text recognizer that not only encodes the input-output attention, but also learns self-Attention which encodes feature-feature and target-target relationships inside the encoder and decoder and learns a more powerful and robust intermediate representation to spatial distortion.
- Regularized vector field learning with sparse approximation for mismatch removal193
This paper proposes a sparse approximation to a robust vector field learning method, sparse vector field consensus (SparseVFC), and derives a statistical learning bound on the speed of the convergence, and applies SparseVFC to the mismatch removal problem.
- Shape matching and classification using height functions193
A novel shape descriptor for matching and recognizing 2D object silhouettes that is not only invariant to geometric transformations such as translation, rotation and scaling but also insensitive to nonlinear deformations due to noise and occlusion is proposed.
- PRA-Net: Point Relation-Aware Network for 3D Point Cloud Analysis189
A novel framework named Point Relation-Aware Network (PRA-Net), which is composed of an Intra-region Structure Learning (ISL) module and an Inter-region Relation Learning (IRL) modules that can dynamically integrate the local structural information into the point features, and captures inter-region relations adaptively and efficiently.
- EPNet++: Cascade Bi-Directional Fusion for Multi-Modal 3D Object Detection187
EPNet++ is proposed for multi-modal 3D object detection by introducing a novel Cascade Bi-directional Fusion (CB-Fusion) module and a Multi-Modal Consistency (MC) loss, which explicitly guarantees the consistency between predicted scores from two modalities.
- Deep Learning Representation using Autoencoder for 3D Shape Retrieval187
This work projects 3D shapes into 2D space and uses autoencoder for feature learning on the 2D images and shows the proposed deep learning feature is complementary to conventional local image descriptors, which can obtain the state-of-the-art performance on 3D shape retrieval benchmarks.
- Scene Text Image Super-Resolution in the Wild186
A new Text Super-Resolution Network termed TSRN, with three novel modules is developed, which improves the recognition accuracy by over 13% of CRNN, and by nearly 9.0% of ASTER and MORAN compared to synthetic SR data.
- 185
- Bag of contour fragments for robust shape classification178
A new shape representation called Bag of Contour Fragments (BCF) inspired by classical Bag of Words (BoW) model is developed, which achieves the state-of-the-art performance on several well-known shape benchmarks, and can be applied to real image classification problem.
- Editing Text in the Wild171
This work proposes an end-to-end trainable style retention network (SRNet) that consists of three modules: text conversion module, background inpainting module and fusion module, which is the first attempt to edit text in natural images at the word level.
- 168
- SegLink++: Detecting Dense and Arbitrary-shaped Scene Text by Instance-aware Component Grouping164
A network for detecting dense and arbitrary-shaped scene text by instance-aware component grouping (ICG), which is a flexible bottom-up method that significantly outperforms state-of-the-art methods on DAST1500 and two curved text datasets.
- SeqFormer: Sequential Transformer for Video Instance Segmentation163
This work presents SeqFormer, a model that locates an instance in each frame and aggregates temporal information to learn a powerful representation of a video-level instance, which is used to predict the mask sequences on each frame dynamically.
- SparseTrack: Multi-Object Tracking by Performing Scene Decomposition Based on Pseudo-Depth162
A pseudo-depth estimation method for obtaining the relative depth of targets from 2D images and a depth cascading matching (DCM) algorithm, which can use the obtained depth information to convert a dense target set into multiple sparse target subsets and perform data association on these sparse target subsets in order from near to far are proposed.
- TextScanner: Reading Characters in Order for Robust Scene Text Recognition162
TextScanner bears three characteristics: it belongs to the semantic segmentation family, as it generates pixel-wise, multi-channel segmentation maps for character class, position and order, and also adopts RNN for context modeling.
- 158
- Improving Shape Retrieval by Learning Graph Transduction155
This paper considers the existing shapes as a group, and study their similarity measures to the query shape in a graph structure, and learns a better metric through graph transduction by propagating the model through existing shapes, in a way similar to com- puting geodesics in shape manifold.
- Symmetry-Constrained Rectification Network for Scene Text Recognition153
A Symmetry-constrained Rectification Network (ScRN) based on local attributes of text instances, such as center line, scale and orientation, which enables ScRN to generate better rectification results than existing methods and thus lead to higher recognition accuracy.
- Learn to Scale: Generating Multipolar Normalized Density Maps for Crowd Counting153
First, a patch-level density map is extracted by a density estimation model and further grouped into several density levels which are determined over full datasets, which can significantly condense the density distribution into several clusters, and enable that the density variance can be learned by a single model.
- Auto-Encoder Guided GAN for Chinese Calligraphy Synthesis152
This work treats the calligraphy synthesis problem as an image-to-image translation problem and proposes a deep neural network based model which can generate calligraphY images from standard font images directly.
- 150
- All You Need Is Boundary: Toward Arbitrary-Shaped Text Spotting147
This work localizes a set of points on the boundary of each text instance to establish a simple yet effective scheme for end-to-end text spotting, which can read the text of arbitrary shapes.
- Hard-Aware Point-to-Set Deep Metric for Person Re-identification146
In addition to person re-ID, the proposed Hard-Aware Point-to-Set (HAP2S) loss with a soft hard-mining scheme applies to generic deep metric learning benchmarks including CUB-200-2011 and Cars196, and achieves state-of-the-art results.
- 139
- AutoScale: Learning to Scale for Crowd Counting137
A simple and effective Learning to Scale (L2S) module, which automatically scales dense regions into reasonable closeness levels (reflecting image-plane distance between neighboring people), and introduces a customized dynamic cross-entropy loss, significantly improving the localization-based model optimization.
- Action recognition for depth video using multi-view dynamic images130
This study extends dynamic imaging to the depth domain and proposes a novel CNN model to perform feature learning on multi-view dynamic images aimed at enhancing the tuning effectiveness on shallow convolutional layers by alleviating the gradient vanishing problem.
- Script identification in the wild via discriminative convolutional neural network128
The proposed DiscCNN achieves state-of-the-art performances on scene, video and document scripts as well, not requiring any preprocess like binarization, segmentation or hand-crafted features.
- 123
- Cascaded SR-GAN for Scale-Adaptive Low Resolution Person Re-identification118
A unified formulation, a Cascaded Super-Resolution GAN (CSR-GAN) framework is proposed, which requires a network not only synthesizing high-resolution images with different upscaling factors, but also extracting discriminative image feature for judging person’s identity.
- Image stitching by line-guided local warping with global similarity constraint118
A line-guided local warping method with a global similarity constraint for image stitching that provides a convincing stitching performance and that it outperforms other state-of-the-art methods.
- Max-Margin Multiple-Instance Dictionary Learning117
This paper designs a formulation to simultaneously learn mixtures of codes by maximizing classification margins in MIL, and results are observed in image classification benchmarks based on the learned codebooks, which observe both compactness and effectiveness.
- Crop segmentation from images by morphology modeling in the CIE L*a*b* color space117
A new morphology modeling method is utilized to establish the crop color model in the CIE L^*a^*b^* (or Lab for simplification) color space and to realize the crop image segmentation, demonstrating that this method is robust to the variation of illumination in the field and performed better than eight other approaches.
- CrowdCLIP: Unsupervised Crowd Counting via Vision-Language Model115
This work proposes a novel unsupervised framework for crowd counting, named CrowdCLIP, which is the first to investigate the vision-language knowledge to solve the counting problem and surpasses some pop-ular fully-supervised methods under the cross-dataset setting.
- Few Could Be Better Than All: Feature Sampling and Grouping for Scene Text Detection113
This paper presents a simple yet effective transformer-based architecture for scene text detection based on a few representative features, which avoids the disturbance by background and reduces the computational cost.
- Text/non-text image classification in the wild with convolutional neural networks113
A novel convolutional neural network variant, called multi-scale spatial partition network (MSP-Net), which classifies images that contain text or not, by predicting text existence in all image blocks, which are spatial partitions at multiple scales on an input image.
- 110
- 109
- DeepSkeleton: Learning Multi-Task Scale-Associated Deep Side Outputs for Object Skeleton Extraction in Natural Images109
This paper presents a novel fully convolutional network with multiple scale-associated side outputs to address object skeleton extraction from natural images, and achieves promising results on two skeleton extraction datasets, and significantly outperforms other competitors.
- Syntax-Aware Network for Handwritten Mathematical Expression Recognition108
A simple and efficient method is proposed, which is the first to incorporate syntax information into an encoder-decoder network, and can effectively describe the syntax context of expressions, alleviating the structure prediction errors of HMER.
- Object Skeleton Extraction in Natural Images by Fusing Scale-Associated Deep Side Outputs107
A fully convolutional network with multiple scale-associated side outputs is presented to address object skeleton extraction in natural images, and achieves promising results on two skeleton extraction datasets, and significantly outperforms other competitors.
- MOST: A Multi-Oriented Scene Text Detector with Localization Refinement105
A new algorithm for scene text detection is proposed, which puts forward a set of strategies to significantly improve the quality of text localization and achieves state-of-the-art or competitive performance on various standard benchmarks for text detection while keeping a fast running speed.
- 104
- Semantically Multi-Modal Image Synthesis104
A novel Group Decreasing Network (GroupDNet) is proposed that leverages group convolution in the generator and progressively decreases the group numbers of the convolutions in the decoder, which is armed with much more controllability on translating semantic labels to natural images and has plausible high-quality yields for datasets with many classes.
- Multi-shot Temporal Event Localization: a Benchmark102
This paper proposes a new and challenging task called multi-shot temporal event localization, and accordingly, collects a large-scale dataset called MUlti-Shot EventS (MUSES), which has 31,477 event instances for a total of 716 video hours.
- 101
- Skeleton growing and pruning with bending potential ratio101
A novel significance measure for skeleton pruning, called bending potential ratio (BPR), in which the decision regarding whether a skeletal branch should be pruned or not is based on the context of the boundary segment that corresponds to the branch, is proposed.
- SPTS v2: Single-Point Scene Text Spotting99
The SPTS v2 can outperform previous state-of-the-art single-point text spotters with fewer parameters while achieving 19× faster inference speed and experiments suggest a potential preference for single- point representation in scene text spotting when compared to other representations.
- 99
- Turning a CLIP Model into a Scene Text Detector96
This paper proposes a new method, termed TCM, focusing on Turning the CLIP Model directly for text detection without pretraining process, and achieves promising domain adaptation ability.
- Divide and Fuse: A Re-ranking Approach for Person Re-identification96
A "Divide and use" re- ranking framework for person re-ID that exploits the diversity from different parts of a high-dimensional feature vector for fusion-based re-ranking, while no other features are accessible.
- 96
- 91
- 91
- 90
- Similarity Fusion for Visual Tracking88
The approach contributes in two different aspects: multiple features are integrated into a unified similarity to enhance the discriminative ability of similarity measurements, and the neighborhood context of the samples in forthcoming frame are employed to further improve the measurements.
- Dynamic Adapter Meets Prompt Tuning: Parameter-Efficient Transfer Learning for Point Cloud Analysis86
This paper freezes the parameters of the default pretrained models and proposes the Dynamic Adapter, which generates a dynamic scale for each token, considering the token significance to the downstream task, and seamlessly integrates Dynamic Adapter with Prompt Tuning (DAPT) by constructing Internal Prompts.
- 86
- DISC-FinLLM: A Chinese Financial Large Language Model based on Multiple Experts Fine-tuning85
This work proposes Multiple Experts Fine-tuning Framework, a methodology that improves general LLMs by endowing them with multi-turn question answering abilities, domain text processing capabilities, mathematical computation skills, and retrieval-enhanced generation capabilities.
- WDNet: Watermark-Decomposition Network for Visible Watermark Removal85
This work combines traditional watermarked image decomposition into a two-stage generator, called Watermark-Decomposition Network (WDNet), where the first stage predicts a rough decomposition from the whole watermark image and the second stage specifically centers on the watermarked area to refine the removal results.
- 85
- SPTS: Single-Point Text Spotting84
This work demonstrates that training scene text spotting models can be achieved with an extremely low-cost annotation of a single-point for each instance, and shows that the performance is not very sensitive to the positions of the point annotation, meaning that it can be much easier to be annotated or even be automatically generated than the bounding box that requires precise positions.
- Advancing COVID-19 diagnosis with privacy-preserving collaboration in artificial intelligence82
A privacy-preserving AI framework for CT-based COVID-19 diagnosis is presented and demonstrated on data from 23 hospitals in China and the United Kingdom, showing that the federated learning framework model considerably outperformed all of the local models.
- Learning Sparse and Identity-Preserved Hidden Attributes for Person Re-Identification81
This work proposes to incrementally generate Deep Hidden Attribute (DHA) based on baseline deep network for newly uncovered annotations based on an auto-encoder model that can be plugged into any deep network to mine latent information in an unsupervised manner.
- Integrating Scene Text and Visual Appearance for Fine-Grained Image Classification81
The main idea is combining word representations and deep visual features in a globally trainable deep convolutional neural network for fine-grained image classification, which significantly outperforms classification with only visual representation.
- 80
- 80
- Learning context-sensitive similarity by shortest path propagation80
A novel shape/object retrieval algorithm shortest path propagation (SSP) given a query object q and a target database object p, which explicitly finds the shortest path between them in the distance manifold of the database objects.
- 79
- Image Caption Generation with Part of Speech Guidance78
Experimental results on the most popular benchmark datasets, e.g., Flickr30k and MS COCO, consistently demonstrate that the method can significantly enhance the performance of a standard image caption generation model, and achieve the conpetitive results.
- An Empirical Study of End-to-End Temporal Action Detection77
A mid-resolution baseline detector is built, which achieves the state-of-the-art performance of end-to-end temporal action detection methods while running more than 4×faster.
- 73
- SAM3D: zero-shot 3D object detection via the segment anything model72
This paper proposes a SAM-powered BEV processing pipeline to detect objects and gets promising results on the large-scale Waymo open dataset and presents the opportunity to unleash their power on 3D vision tasks.
- 71
- SOOD: Towards Semi-Supervised Oriented Object Detection70
A novel Semi-supervised Oriented Object Detection model, termed SOOD, built upon the mainstream pseudo-labeling framework is proposed, which surpasses the state-of-the-art SSOD methods under various settings on the DOTAv1.5 benchmark.
- Regularized Diffusion Process for Visual Retrieval70
A new variant o diffusion process, which also operates on a tensor product graph is proposed, defined in three equivalent formulations (regularization framework, iterative framework and limit framework, respectively), which theoretically explain how this kind of diffusion process can better reveal the intrinsic relationship between objects.
- Smart Electronic Nose Enabled by an All‐Feature Olfactory Algorithm68
A deep‐learning‐enabled codesign of sensor arrays and recognition algorithms can reduce the heavy demand for a huge amount of highly specialized gas sensors and provide interpretable insights into odor recognition dynamics in an iterative way.
- Detection and recognition of contour parts based on shape similarity68
A system for detection and recognition of contour parts in digital images based on shape similarity to retrieve the most similar contours parts in a database of known contour segments.
- View N-Gram Network for 3D Object Retrieval66
Inspired by n-gram models in natural language processing, VNN divides the view sequence into a set of visual n- grams, which involve overlapping consecutive view sub-sequences, which helps to learn a discriminative global embedding for each 3D object.
- When Counting Meets HMER: Counting-Aware Network for Handwritten Mathematical Expression Recognition61
- CAPE: Camera View Position Embedding for Multi-View 3D Object Detection60
This paper proposes a novel method based on CAmera view Position Embedding, called CAPE, which form the 3D position embed-dings under the local camera-view coordinate system instead of the global coordinate system, such that3D position em-bedding is free of encoding camera extrinsic parameters.
- Automatic script identification in the wild57
A large-scale dataset with a great quantity of natural images and 10 types of widely-used languages is constructed and a deep learning based algorithm is proposed for script identification at word or line levels in natural scenes.
- Multiple instance subspace learning via partial random projection tree for local reflection symmetry in natural images57
A novel multiple instance learning framework for local reflection symmetry detection, named multiple instance subspace learning (MISL), which instead learns a group of models respectively on well partitioned subspaces, and an efficient dividing strategy under MIL setting, named partial random projection tree (PRPT).
- A Deep End-to-End Model for Transient Stability Assessment With PMU Data55
The results of experiments on IEEE 50-machine system demonstrate the superiority of the proposed model over the prior methods for TSA, which is a unified framework integrating multiple stacked denoising autoencoders (SDAEs), one fusion layer, and one logistic regression (LR) layer.
- ESTextSpotter: Towards Better Scene Text Spotting with Explicit Synergy in Transformer54
This paper introduces a new model named Explicit Synergy-based Text Spotting Transformer framework (ESTextSpotter), which achieves explicit synergy by modeling discriminative and interactive features for text detection and recognition within a single decoder.
- DeepFlux for Skeletons in the Wild54
This article training a CNN to predict a two-dimensional vector field, which maps each scene point to a candidate skeleton pixel, in the spirit of flux-based skeletonization algorithms, achieves consistently superior performance over state-of-the-art methods.
- FaceController: Controllable Attribute Editing for Face in the Wild53
This work proposes a simple feed-forward network to generate high-fidelity manipulated faces with one or multiple desired face attributes manipulated while other details are preserved by simply employing some existing and easy-obtainable prior information.
- SynthText3D: synthesizing scene text images from 3D virtual worlds53
This paper proposes to synthesize scene text images from the 3D virtual worlds, where the precise descriptions of scenes, editable illumination/visibility, and realistic physics are provided.
- AutoSTR: Efficient Backbone Search for Scene Text Recognition53
This work designs a domain-specific search space for STR, which contains both choices on operations and constraints on the downsampling path, and proposes a two-step search algorithm, which decouples operations and downsampled path, for an efficient search in the given space.
- Discrete Skeleton Evolution53
The proposed pruning method allows us to overcome the instability of skeleton representation: a small boundary deformation leads to large changes in skeleton topology and is able to obtain very stable skeleton representation of planar shapes.
- Large-Scale 3D Shape Retrieval from ShapeNet Core5552
This track aims to provide a benchmark to evaluate large-scale shape retrieval based on the ShapeNet dataset, using ShapeNet Core55, which provides more than 50 thousands models over 55 common categories in total for training and evaluating several algorithms.
- Skeleton pruning as trade-off between skeleton simplicity and reconstruction error52
This work proposes a simple algorithm to approximate the maximum of the Bayesian posterior probability which defines an order for iteratively removing the end branches to obtain the pruned skeleton and presents experimental results obtained without any parameter tuning.
- 52
- 51
- 50
- 50
- A collaborative online AI engine for CT-based COVID-19 diagnosis50
A federated learning-based Unified CT-COVID AI Diagnostic Initiative (UCADI), a decentralized architecture where the AI model is distributed to and executed at each host institution with the data sources or client ends for training and inferencing without sharing individual patient data is proposed.
- A Discrepancy Aware Framework for Robust Anomaly Detection49
A discrepancy aware framework (DAF) is presented, which demonstrates robust performance consistently with simple and cheap strategies across different anomaly detection benchmarks and leverages an appearance-agnostic cue to guide the decoder in identifying defects, thereby alleviating its reliance on synthetic appearance.
- Shape clustering: Common structure discovery49
This paper aims to address the problem of shape clustering by discovering the common structure which captures the intrinsic structural information of shapes belonging to the same cluster, based on a skeleton graph, named common structure skeleton graph (CSSG), which expresses possible correspondences between nodes of the individual skeletons of the cluster.
- VD-SAN: Visual-Densely Semantic Attention Network for Image Caption Generation48
The whole captioning system consists of shared convolutional layers from Dense Convolutional Network (DenseNet), which are further split into a semantic attributes prediction branch and an image feature extraction branch, two semantic attention models, and a long short-term memory networks (LSTM) for caption generation.
- 47
- Deep sketch feature for cross-domain image retrieval47
This paper shows that deep convolutional neural networks (DCNN) is also suitable for cross-domain image recognition, i.e., using sketch as query to retrieve natural images in a large dataset, and proposes to train CNN jointly using image data and sketch data in a novel way.
- 46
- 46
- Scene Text Retrieval via Joint Text Detection and Similarity Learning46
An end-to-end trainable network is established, jointly optimizing the procedures of scene text detection and cross-modal similarity learning, and the proposed framework of joint detection and similarity learning achieves significantly better performance than separated methods.
- Fusion with Diffusion for Robust Visual Tracking46
The advantages of the proposed approach to integrate multiple similarity measures is demonstrated on the task of visual tracking, where different aspects of the appearance similarity between the target object in frame t and target object candidates in frames t are integrated.
- 44
- Beyond diffusion process: Neighbor set similarity for fast re-ranking44
A simple yet effective method called Neighbor Set Similarity (NSS) is proposed, which is superior to both traditional pairwise similarity and diffusion process, and suggests that NSS is more suitable for large scale image retrieval than diffusion process.
- 43
- 43
- Class-Aware Mask-guided feature refinement for scene text recognition42
This paper introduces canonical class-aware glyph masks generated from a standard font to effectively suppress background and text style noise, thereby enhancing feature discrimination and highlighting the importance of incorporating canonical mask guidance and aligned feature refinement techniques for robust scene text recognition.
- Vision-Language Pre-Training for Boosting Scene Text Detectors42
This paper proposes to learn contextualized, joint representations through vision-language pretraining, for the sake of enhancing the performance of scene text detectors, and devise a pre-training architecture with an image encoder, a text encoder and a cross-modal encoder.
- A Large Cross-Modal Video Retrieval Dataset with Reading Comprehension40
A large-scale and cross-modal Video Retrieval dataset with text reading comprehension, TextVR, which contains 42.2k sentence queries for 10.5k videos of 8 scenario domains and designs a novel multimodal video retrieval baseline for the text-based video retrieval task.
- 40
- 38
- 38
- 37
- SKELETON-BASED SHAPE CLASSIFICATION USING PATH SIMILARITY37
Using a new skeleton pruning method, it is able to obtain stable pruned skeletons even in the presence of significant contour distortions, and a new method for matching of skeleton graphs is proposed.
- Dynamic Multi-Task Learning with Convolutional Neural Network35
This paper considers the integration of CNN and multi-task learning in a novel way to further improve the performance of multiple related tasks and proposes a dynamic multi-task CNN model that outperforms traditional approaches.
- 34
- 34
- 34
- Affinity Feature Strengthening for Accurate, Complete and Robust Vessel Segmentation33
A novel approach, the affinity feature strengthening network (AFN), which jointly models geometry and refines pixel-wise segmentation features using a contrast-insensitive, multiscale affinity approach and outperforms the state-of-the-art methods in both higher accuracy and topological metrics.
- Smooth Neighborhood Structure Mining on Multiple Affinity Graphs with Applications to Context-Sensitive Similarity33
A new algorithm called Smooth Neighborhood (SN) is proposed that mines the neighborhood structure to satisfy the manifold assumption and is adjusted to tackle multiple affinity graphs by imposing a weight learning paradigm.
- 30
- 30
- Modeling Entities as Semantic Points for Visual Information Extraction in the Wild29
Extensive experiments on standard benchmarks in this field as well as the proposed dataset demonstrate that the proposed method can achieve significantly enhanced performance on entity labeling and linking, compared with previous state-of-the-art models.
- 29
- 29
- 28
- 28
- 28
- 28
- 28
- Neural shape codes for 3D model retrieval28
This work proposes a practical and effective way for applying Convolutional Neural Network to 3D model retrieval, by training the network with the depth projections of3D model.
- 28
- 28
- 26
- 26
- 26
- 25
- 25
- 25
- 25
- 25
- OmniParser V2: Structured-Points-of-Thought for Unified Visual Text Parsing and Its Generality to Multimodal Large Language Models24
OmniParser V2 is introduced, a universal model that unifies VsTP typical tasks, including text spotting, key information extraction, table recognition, and layout analysis, into a unified framework, and central to the proposed Structured-Points-of-Thought prompting schemas, which improves model performance across diverse scenarios.
- 24
- 24
- 23
- 23
- 22
- 22
- Incremental Deep Hidden Attribute Learning22
This work proposes a new person representation with hidden attributes by mining latent information from visual feature in an unsupervised manner using an auto-encoder model plugged-in to the deep learning network to compose a Deep Hidden Attribute Network (DHA-Net).
- Multidimensional Scaling on Multiple Input Distance Matrices22
This paper proposes a new algorithm called Multi-View Multidimensional Scaling (MVMDS) by considering each input distance matrix as one view, which can learn the weights of views automatically by exploring the consensus information and complementary nature of views.
- 22
- 21
- 21
- 21
- 20
- 20
- 20
- 20
- 19
- 19
- 19
- 19
- 18
- Feature context learning for human parsing18
This paper proposes an end-to-end trainable, highly flexible and generic module called feature context module (FCM), which explores the correlation of adjacent pixels and aggregates the contextual information embedded in the real topology of the human body.
- 17
- 17
- 16
- 16
- 16
- 15
- 15
- 15
- 15
- 15
- 15
- 14
- 14
- 14
- 14
- 2D Scene Sketch-Based 3D Scene Retrieval14
The objective of this track is to evaluate the performance of different 2D scene sketch-based 3D scene retrieval algorithms using a 2D sketch query dataset and a 3D Warehouse model dataset and suggest several future research directions to tackle this research problem.
- 13
- 13
- High-Resolution Photo Enhancement in Real-Time: A Laplacian Pyramid Network12
A pyramid network called LLF-LUT++ is introduced, which integrates global and local operators through closed-form Laplacian pyramid decomposition and reconstruction, which enables fast processing of high-resolution images while also achieving excellent performance.
- 12
- Searching a High Performance Feature Extractor for Text Recognition Network12
This article designs a domain-specific search space by exploring principles for having good feature extractors in a huge and complexly structured space and proposes a two-stage algorithm to effectively search in the space.
- Improving context-sensitive similarity via smooth neighborhood for object retrieval12
A new algorithm called Smooth Neighborhood (SN) is proposed that mines the neighborhood structure to satisfy the manifold assumption and can yield state-of-the-art performances on shape retrieval, image retrieval and 3D model retrieval.
- Multiple Stage Residual Model for Accurate Image Classification12
This paper proposes an effective and efficient model, which is called Multiple Stage Residual Model (MSRM), to make full use of the residual vector to generate a multiple stage code.
- 11
- 11
- 11
- 11
- 11
- A comparison of methods for 3D scene shape retrieval11
A comprehensive comparison of all the above methods and an additional new retrieval methods are performed by evaluating them on the two benchmarks to solicit state-of-the-art approaches.
- 11
- 10
- 10
- 10
- 10
- 10
- 10
- 10
- 9
- 9
- InstMove: Instance Motion for Object-centric Video Segmentation9
InstMove, which stands for Instance Motion for Object-centric Video Segmentation, mainly relies on instance-level motion information that is free from image feature embeddings, and features physical interpretations, making it more accurate and robust toward occlusion and fast-moving objects.
- 9
- 9
- 9
- 9
- 8
- 8
- 8
- 8
- 8
- 8
- 8
- 8
- 7
- CauESC: A Causal Aware Model for Emotional Support Conversation7
A novel framework CauESC is proposed, which firstly recognizes the emotion causes of the distress, as well as the emotion effects triggered by the causes, and then understands each strategy of verbal grooming independently and integrates them skillfully.
- 7
- 7
- 7
- 7
- 6
- Describe, Adapt and Combine: Empowering CLIP Encoders for Open-Set 3D Object Retrieval6
A simple yet effective framework named Describe, Adapt and Combine (DAC) is presented, which innovatively synergizes a CLIP model with a multi-modal large language model (MLLM) to learn generalized 3D representations, where the MLLM is used for dual purposes.
- 6
- 6
- Deep Interactive Video Inpainting: An Invisibility Cloak for Harry Potter6
This is the first deep learning-based interactive video inpainting framework that only uses a free form of user input as guidance (i.e. scribbles) instead of mask annotations, which has academic, entertainment, and commercial value.
- 6
- 6
- 6
- 6
- 6
- 6
- 5
- 5
- VimTS: A Unified Video and Image Text Spotter for Enhancing the Cross-Domain Generalization5
This paper introduces a new method, termed VimTS, which enhances the generalization ability of the model by achieving better synergy among different tasks, and proposes a synthetic video text dataset (VTD-368 k) by leveraging the Content Deformation Fields (CoDeF) algorithm.
- 5
- 5
- 5
- 5
- Computer Vision5
The idea is to randomize attribute selection and no longer look for globally optimal split, but instead randomly use subset of K attributes on which to base the split.
- 5
- Co-spectral for robust shape clustering5
A similarity fusion framework based on spectral analysis is proposed, which can automatically make use of the complementarity of various shape similarities based on a co-training framework and outperforms the state-of-the-art algorithms by a large margin.
- 5
- 5
- 5
- 5
- 5
- 4
- 4
- 4
- 4
- 4
- 4
- 4
- 4
- 4
- 4
- 4
- Generative compositor for few-shot visual information extraction3
A novel generative model, named Generative Compositor, is proposed to address the challenge of few-shot VIE, a hybrid pointer-generator network that emulates the operations of a compositor by retrieving words from the source text and assembling them based on the provided prompts.
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- Ceria supported nickel nanoparticles for catalytic hydroconversion of lignin-related model compounds3
- 3
- 3
- 3
- 3
- 3
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- The Runner-up Solution for YouTube-VIS Long Video Challenge 2022–
This technical report describes the 2nd-place solution for the ECCV 2022 YouTube-VIS Long Video Challenge, which adopts the previously proposed online video instance segmentation method IDOL and uses pseudo labels to help contrastive learning.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
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 2026-10-13. 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.