Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks134 from public data
- Dataset Condensation with Distribution Matching507
This work proposes a simple yet effective method that synthesizes condensed images by matching feature distributions of the synthetic and original training images in many sampled embedding spaces and significantly reduces the synthesis cost while achieving comparable or better performance.
- 357
- A survey on deep learning-based fine-grained object classification and semantic segmentation337
Four types of deep learning based fine-grained image classification approaches are reviewed, including the general convolutional neural networks (CNNs), part detection based, ensemble of networks based and visual attention based fine -grainedimage classification approaches.
- Review of Weed Detection Methods Based on Computer Vision303
This review elaborates the two aspects of using traditional image-processing methods and deep learning-based methods to solve weed detection problems and provides an overview of various methods for weed detection in recent years.
- MLVU: Benchmarking Multi-task Long Video Understanding294
A new benchmark called MLVU (Multitask Long Video Understanding Benchmark) is proposed for the comprehensive and in-depth evaluation of LVU, which suggests that factors such as context length, image-understanding ability, and the choice of LLM backbone can play critical roles in future advancements.
- Image Generation From Layout245
The proposed Layout2Im model significantly outperforms the previous state of the art, boosting the best reported inception score by 24.66% and 28.57% on the very challenging COCO-Stuff and Visual Genome datasets, respectively.
- SpatialBot: Precise Spatial Understanding with Vision Language Models220
SpatialBot, a model designed to enhance spatial understanding by utilizing both RGB and depth images, is proposed and the SpatialQA and SpatialQA $\boldsymbol{E}$ datasets are introduced to train VLMs for depth perception.
- Multi-View Image Generation from a Single-View193
This paper proposes a novel image generation model termed VariGANs, which combines the merits of the variational inference and the Generative Adversarial Networks (GANs), and generates the target image in a coarse-to-fine manner instead of a single pass which suffers from severe artifacts.
- Neural Person Search Machines171
Evaluations on two benchmark datasets have demonstrated that the Neural Person Search Machines (NPSM) developed can outperform current state-of-the-arts in both mAP and top-1 evaluation protocols.
- CAFE: Learning to Condense Dataset by Aligning Features170
This paper proposes a novel scheme to Condense dataset by Aligning FEatures (CAFE), which explicitly attempts to preserve the real-feature distribution as well as the discriminant power of the resulting synthetic set, lending itself to strong generalization capability to various architectures.
- 150
- 124
- Development of a tomato harvesting robot used in greenhouse111
Some insights are provided into the development and application of tomato harvesting robot used in the greenhouse and obstacle avoidance strategies proposed based on the C-space method.
- Enhancement of Detecting Permanent Water and Temporary Water in Flood Disasters by Fusing Sentinel-1 and Sentinel-2 Imagery Using Deep Learning Algorithms: Demonstration of Sen1Floods11 Benchmark Datasets109
This paper proposes an automatic segmentation method for surface water, permanent water, and temporary water identification, and all tasks share the same convolutional neural network architecture and utilizes focal loss to deal with the class (water/non-water) imbalance problem.
- 101
- Interpretable Spatio-Temporal Attention for Video Action Recognition97
This model not only improves video action recognition accuracy, but also localizes discriminative regions both spatially and temporally, despite being trained in a weakly-supervised manner with only classification labels (no bounding box labels or time frame temporal labels).
- Left-Right Comparative Recurrent Model for Stereo Matching89
A novel left- right comparative recurrent model to perform left-right consistency checking jointly with disparity estimation and a soft attention mechanism is introduced, which employs the learned error maps for better guiding the model to selectively focus on refining the unreliable regions at the next recurrent step.
- CT synthesis from MR in the pelvic area using Residual Transformer Conditional GAN85
A new GAN called Residual Transformer Conditional GAN (RTCGAN), which exploits the advantages of CNN in local texture details and Transformer in global correlation to extract multi-level features from MR and CT images and achieves desirable results on local mismatch tissues.
- Weed and Corn Seedling Detection in Field Based on Multi Feature Fusion and Support Vector Machine82
The experimental results showed that the fusion feature combination of rotation invariant LBP feature and gray level-gradient co-occurrence matrix based on SVM classifier obtained the highest classification accuracy and accurately detected all kinds of weeds and corn seedlings.
- Modular Generative Adversarial Networks82
ModularGAN consists of several reusable and composable modules that carry on different functions that can be trained simultaneously, leveraging data from all domains, and then combined to construct specific GAN networks at test time, according to the specific image translation task.
- Application of Convolutional Neural Network-Based Detection Methods in Fresh Fruit Production: A Comprehensive Review80
Improved CNN deep learning models can give full play to detection potential by combining with the characteristics of each link of fruit production, and imply that CNN-based detection may penetrate the challenges created by environmental issues, new area exploration, and multiple task execution of fresh fruit production in the future.
- Automatic generation of large-scale handwriting fonts via style learning79
Using the proposed system, for the first time the practical handwriting font library in a user's personal style with arbitrarily large numbers of Chinese characters can be generated automatically.
- 73
- Minimal mask immobilization with optical surface guidance for head and neck radiotherapy61
These pilot results confirm feasibility of minimal mask immobilization combined with commercially available optical image guidance and follow‐up validation, with direct comparison to standardMask immobilization, appears warranted.
- EasyFont59
Using the proposed EasyFont system, for the first time, the practical handwriting font library in a user’s personal style with arbitrarily large numbers of Chinese characters can be generated automatically.
- Continual Representation Learning for Biometric Identification55
This paper proposes a new continual learning setting, namely "continual representation learning", which focuses on learning better representation in a continuous way, and demonstrates that existing CL methods can improve the representation in the new setting.
- A Large-Scale Attribute Dataset for Zero-Shot Learning55
A Large-scale Attribute Dataset with 78,017 images of 230 classes with 359 attributes of visual, semantic and subjective properties is proposed, which argues that the "co-occurrence bias problem" of existing datasets, which is caused by the biased co-occurring of objects, significantly hinders models from correctly learning the concept.
- 49
- Layout2image: Image Generation from Layout47
The proposed Layout2Im model significantly outperforms the previous state-of-the-art, boosting the best reported inception score by 24.66% and 28.57% on the very challenging COCO-Stuff and Visual Genome datasets, respectively.
- Sketch Recognition with Deep Visual-Sequential Fusion Model39
A deep end-to-end network for sketch recognition, named Deep Visual-Sequential Fusion model (DVSF) is proposed to model the visual and sequential patterns of the strokes, which outperforms the state-of-the-art approaches.
- 37
- Zero-Shot Learning Posed as a Missing Data Problem36
This paper presents a method of zero-shot learning (ZSL) which poses ZSL as the missing data problem, rather than the missing label problem, and explores a simple yet effective transductive framework in the reverse way.
- Automatic segmentation for brain MR images via a convex optimized segmentation and bias field correction coupled model36
A novel automatic variational method, which can segment brain MR images meanwhile correcting the bias field when segmenting images with high intensity inhomogeneities and is able to estimate the bias of quite general profiles, even in 7T MR images.
- 35
- Unveiling the Ignorance of MLLMs: Seeing Clearly, Answering Incorrectly34
This investigation identifies two primary issues: 1) most instruction tuning datasets predominantly feature questions that "directly" relate to the visual content, leading to a bias in MLLMs’ responses to other indirect questions, and 2) MLLMs’ attention to visual tokens is notably lower than to system and question tokens.
- 33
- Initial clinical experience with a radiation oncology dedicated open 1.0T MR‐simulation29
Clinical applicability was demonstrated for 4D MRI and UTE imaging to support MR‐SIM for single modality treatment planning and staffing, safety, and patient screening procedures were developed.
- 28
- Learning fashion compatibility across categories with deep multimodal neural networks23
A novel multimodal framework to learn fashion compatibility, which simultaneously integrates both semantic and visual embeddings into a unified deep learning model, which outperforms the state-of-the-art approaches.
- Technical Solution Discussion for Key Challenges of Operational Convolutional Neural Network-Based Building-Damage Assessment from Satellite Imagery: Perspective from Benchmark xBD Dataset22
Four specific data resampling strategies are provided, which are Main-Label Over-Sampling (MLOS), Discrimination After Cropping (DAC), Dilation of Area with Minority (DAM) and Synthetic Minority Over- Sampling Technique (SMOTE), as well as cost-sensitive re-weighting schemes.
- Orthogonal curved‐line Gabor filter for fast fingerprint enhancement22
This paper presents an orthogonal curved-line Gabor filter, which decomposes the unrotated two-dimensional (2D) Gabor into a 1D bandpass Gabor and a1D low pass Gabor, and implements them with two separable steps: a 2D band pass and 1D lowpass filtering on curved lines, respectively, perpendicular and parallel to the ridge orientation.
- 21
- 19
- 19
- 18
- CT synthesis from MR images using frequency attention conditional generative adversarial network17
A frequency cycle generative model (FCGM) is designed to enhance the inter-mapping between MR and CT and extract more rich tissue structure information and produces higher-quality sCT images while retaining clearer and richer high-frequency texture information.
- Detection Method for Rice Seedling Planting Conditions Based on Image Processing and an Improved YOLOv8n Model15
This research provides data support for the efficiency and quality of transplanting machine operations, contributing to the further development and application of unmanned field management in subsequent rice seedling cultivation.
- 15
- Attribute-Guided Image Generation from Layout15
A new image generation method that enables instance-level attribute control, and the generated images from this model have higher resolution, object classification accuracy and consistency, as compared to the previous state-of-the-art.
- 14
- 14
- Human Action Recognition Based on Deep Learning14
This paper will enumerate the main action recognition network model, and introduce the characteristics of different network models, and provide some valuable reference for the research of action recognition.
- Dosimetric Evaluation of Fractionated Stereotactic Radiation Therapy for Skull Base Meningiomas Using HyperArc and Multicriteria Optimization11
HyperArc radiosurgery optimization and delivery in conjunction with multicriteria optimization (MCO) to create conformal and efficient treatment plans for conventionally fractionated radiation therapy to difficult base-of-skull (BOS) lesions is evaluated.
- Cer-ConvN3Unet: an end-to-end multi-parametric MRI-based pipeline for automated detection and segmentation of cervical cancer10
An AI-assisted multiparametric MRI-based pipeline can effectively support radiologists in cervical cancer evaluation and shows high recognition and segmentation performance without manual intervention, effectively reducing the delineation burden for radiologists and gynecologists.
- TomatoPoseNet: An Efficient Keypoint-Based 6D Pose Estimation Model for Non-Destructive Tomato Harvesting10
The experimental results demonstrate that the proposed TomatoPoseNet method can accurately locate the 6D pose of cutting points in an unstructured tomato harvesting environment, enabling non-destructive harvesting.
- TC-LLaVA: Rethinking the Transfer from Image to Video Understanding with Temporal Considerations10
This paper proposes two strategies to enhance the model's capability in video understanding tasks by improving inter-layer attention computation in LLMs and adapt LLaVA for video understanding tasks, naming it Temporal-Considered LLaVA (TC-LLaVA).
- Dose convolution filter: Incorporating spatial dose information into tissue response modeling10
The DCF model incorporates spatial dose information and offers an improved way to estimate tissue response from complex radiotherapy dose distributions, which does not alter the prediction of tissue response in large homogenous fields, but successfully predicts increased tissue tolerance in small or highly nonuniform fields.
- 9
- 9
- 9
- 9
- Electrical Data Matrix Decomposition in Smart Grid9
Inspired by matrix decomposition, a low-rank electricity data matrix is built where the row is time and the column is user and the multiply of two small matrices is used to approximate the low-rank electricity data matrix and recover the missed electrical data.
- Deep-learning based surface region selection for deep inspiration breath hold (DIBH) monitoring in left breast cancer radiotherapy9
A convolutional neural network based automatic region-of-interest (ROI) selection method was proposed to select an optimal surface ROI for DIBH surface monitoring and the proposed RE predictive model can be utilized for selecting a quasi-optimal ROI in left breast cancer DibH radiotherapy (DIBH-RT).
- Clothing Extraction by Coarse Region Localization and Fine Foreground/Background Estimation9
An automatic clothing extraction algorithm is proposed, which consists of two phases: coarse clothing region localization with human proportion, and fine foreground/background modeling, which achieves good performance, and has competitive performance with the interactive solution.
- 9
- Using deep learning to model the biological dose prediction on bulky lung cancer patients of partial stereotactic ablation radiotherapy8
The study developed a D-B Nestnet model capable of predicting BED distribution accurately for bulky lung cancer patients in P-SABR that enables a quick intuitive evaluation of tumor ablation, modification of the ablation range to improve BED of tumor targets, and quality assessment.
- Channel Attention Networks for Image Translation8
The proposed channel attention networks for image translation is a novel model that supports the multi-domain image-to-image translation using one single model that does not require paired training data.
- 8
- 8
- 7
- 7
- 7
- A parallel-res GRU architecture and its application to road network traffic flow forecasting7
This paper explicitly reformulate the Parallel-Res Path as learning residual functions with reference to the input layer and output layer, instead of stacked plain GRU layers, and constructs multi-level residual architecture.
- 7
- 6
- Normalizing Batch Normalization for Long-Tailed Recognition5
This paper identifies that the bias towards the frequent class may be encoded into features, i.e., the rare-specific features which play a key role in discriminating the rare class are much weaker than the frequent-specific features.
- Identification and expression of the Di19 gene family in response to abiotic stress in common bean (Phaseolus vulgaris L.)5
The most common bean Di19 genes were inhibited or not regulated by cadmium stress, but the expression of PvDi19-6 in roots was significantly upregulated when subjected to lower concentrations of cadmium (5 mmol).
- Multi-Feature Recognition of Healthy Vegetable Seedlings Based on Machine Vision Technology5
An image recognition technology based on machine vision is proposed for the non-destructive detection of healthy vegetable seedlings, aiming to improve the accuracy of identifying healthy vegetable potted seedlings.
- Online Measuring and Size Sorting for Perillae Based on Machine Vision5
An approach based on the machine vision (MV) technique for online measuring and size sorting is developed, including image binarization, key point determination, information matching, and parameter estimation for perillae sizing.
- 5
- Considering marker visibility during leaf sequencing for segmental intensity‐modulated radiation therapy5
A new leaf sequencing algorithm based on the Kamath algorithm that is optimal in terms of MU efficiency and marker visibility is developed and rigorously proved its optimality.
- Touchstone Benchmark: Are We on the Right Way for Evaluating AI Algorithms for Medical Segmentation?4
- 4
- Development of automated region of interest selection algorithms for surface‐guided radiation therapy of breast cancer4
The proposed automated algorithms can improve the ROI compliance with the SGRT protocol, with a shortened preparation time, and is ready to be integrated into the clinical workflow for automated ROI preparation.
- Mask guided diverse face image synthesis4
A novel end-to-end learning framework to generate diverse, realistic and controllable face images guided by face masks that can generate different face images matching the input face mask and by manipulating the face mask, can finely control the generated face image.
- Efficient Object Detection based on Deep Feature Fusion Network4
An efficient object detection method based on deep feature fusing strategy that employs deep convolution neural network to fuse multi-channel input, including colored image, infrared image and motion image, and the backbone network is based on darknet-53.
- 4
- Targeting accuracy at couch kick for a frameless image guided radiosurgery system.4
Positioning accuracy with stereoscopic x-ray imaging over the complete range of couch rotation using anthropomorphic head phantoms shows that Frameless image-guided stereotactic radiosurgery could achieve targeting accuracy similar to that of frame-based systems for high dose treatment of trigeminal neuralgia.
- 4
- Application of new three-dimensional printed tissue compensators in radiotherapy4
The highly personalized 3D printed tissue compensators can provide a positioning support for patients and effectively increase the dose to the skin in radiotherapy.
- 4
- Interpretable Attention Guided Network for Fine-Grained Visual Classification3
This work proposes an Interpretable Attention Guided Network (IAGN) for fine-grained visual classification, which is the first interpretable FGVC method with a competitive performance on several standard FGVC benchmark datasets.
- 3
- Augmented Bi-path Network for Few-shot Learning3
The proposed Augmented Bi-path Network (ABNet) for learning to compare both global and local features on multi-scales is proposed, where the salient patches are extracted and embedded as the local features for every image.
- An Improved Gaussian Mixture Model based on NonLocal Information for Brain MR Images Segmentation3
The improved method coupled segmentation and bias field correction that can manage the bias field while segmenting the image and show that the method can obtain more accurate results and biasField.
- 2
- 2
- Impact of MRI resolution for Linac-based stereotactic radiosurgery2
Significant improvement in target definition and reduced inter-observer variation was observed as the MRI resolution improved, which also improved the quality of SRS plans, and results imply that high resolution 3D MR sequences should be used to minimize potential errors in targetdefinition, and multi-slice 2D sequence should be avoided.
- 2
- 2
- Brain MR Image Segmentation and Bias Correction Using an Improved Gaussian Mixture Model2
A novel model is proposed which combines segmentation with bias correction that can manage the bias field and noise while segmenting the image and can get a good segmentation results and bias field.
- Target Monitoring on Face Detection Based on Improved AdaBoost Algorithm2
This paper proposes use improved AdaBoost algorithm, which is much better than normal AdaBoost, for face detection in Pattern Recognition and Computer Vision.
- High-Order MRF Prior Based Bayesian Deblurring2
This research presents a novel and scalable approach to solve the challenge of integrating NoSQL data stores to manage complex systems.
- Image Segmentation to HSI Model Based on Improved Particle Swarm Optimization2
The image segmentation method based on the improved PSO was proposed, which can effectively overcome shortages which are the slow rate of the particle swarm optimization and the poor segmentation quality by using other algorithms.
- 1
- An error correction ACKLSTM model combining data-driven and deep learning for wind power prediction1
The comparative analysis demonstrates that the hybrid model, which integrates error correction and KAN, outperforms alternative methods, achieving the lowest mean squared error (MSE) of 0.091 ± 0.01 kW, representing a 70% improvement over the Transformer model.
- 1
- 1
- Squeeze-and-excitation blocks embedded YOLO model for fast target detection under poor imaging conditions1
The improved YOLOv3 model whose basic structure of the detector is based on darknet-53, which is an accurate but efficient network for image feature extraction, and Squeeze-and-Excitation (SE) structure is integrated after non-linearity of convolution to collect spatial and channel-wise information within local receptive fields.
- 1
- Quantifying intra-fractional prostate motion trajectory for establishing a new gating strategy: a preliminary study1
As an essential component of the new gating strategy, trajectory quantification makes it possible to automatically discriminate intra-fractional motion patterns; pattern discrimination further allows active guidance for subsequent interventions.
- 1
- A New Bit-Rate Control Algorithm for H.264 on the Basic Unit Layer1
Experimental results demonstrate the proposed algorithm can predict MAD more accurately than the classic G012 algorithm, and the PSNR gain of the proposed method can be up to 0.08db at most.
- 1
- –
- Optimizing lung SBRT delivery: A hybrid approach combining dynamic conformal arc (DCA) and volumetric modulated arc therapy (VMAT) techniques–
The hDCA‐VMAT technique offers a practical, streamlined approach for lung SBRT planning that reduces modulation while maintaining high plan quality and improves delivery efficiency, especially relevant for treating low‐density lung tumors.
- –
- –
- –
- –
- –
- –
- Person Re-identification Based on Multi-scale Network Attention Fusion–
Experiments show that the method makes full use of the information of different depths of the network and the key information of interest, so that the model has strong discriminating ability, and the average accuracy of the proposed model is better than most state-of-the-art algorithms.
- Research on Image Feature Processing Based on FPGA–
The vector comparison method is used to perform feature retrieval and sorting on images when reading data, thereby achieving high-performance integrated image classification and retrieval functions.
- Proton Improves BED Compared With Photon Beam-based Partial Stereotactic Ablative Boost Radiotherapy for Large Non-small Cell Lung Cancer–
Higher BED inside the target could be achieved by using proton beam therapy with similar OARs dose sparing in high-dose region and significantly better OAR’s low dose sparing compared with photon techniques.
- –
- Iris Recognition Technology and Application Research in the Field of Public Security–
The basic concepts and main advantages of iris recognition, as well as the application of well-known iris Recognition products at home and abroad are introduced.
- An Effective Solution for Application Orchestration–
The goal is to solve the problem of existing application orchestration techniques from the root by borrowing Docker’s quick-start feature and the universal characteristics of hybrid cloud technology.
- –
- –
- Geometrically optimized single-isocenter dynamic conformal arcs (GO-SIDCA): an efficient radiosurgery technique for multiple brain metastasis–
GO-SIDCA is the most efficient for radiosurgery of multiple targets and provides the best MU efficiency and comparable plan quality with SIDCA and VMAT, but for cases without good geometry, inversely optimized VMAT might have an advantage.
- –
- –
- A Miniature Handheld Automatic Detected Device for Register–
Experiments indicate that transversal deviation precision and longitudinal deviation precision of this device is 0.01 millimeter, when this automatic detected device for register of single-sheet offset based on image processing is applied.
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.