New methods enhance image quality while preserving individual identity.
― 6 min read
Computer Science - Computer Vision and Pattern Recognition
RSSA new method enhances privacy in federated learning against gradient inversion attacks.
― 8 min read
New method addresses brightness issues in diffusion models for better image generation.
― 6 min read
A new lightweight vision transformer enhances depth estimation for low-power devices.
― 5 min read
A new benchmark assesses continual learning in multimodal language models.
― 6 min read
DrFER offers more accurate facial expression analysis by separating features from identity.
― 6 min read
Exploring the connection between idempotence and perceptual quality in image compression.
― 5 min read
A new framework simplifies animation from images using clicks and short prompts.
― 7 min read
Introducing REPAIR to improve data pairing accuracy in noisy environments.
― 6 min read
A new method improves category discovery by combining visual and text information.
― 7 min read
ProMark offers a method for attributing generated images to their original sources.
― 5 min read
Robots learn to find moving objects in dynamic environments effectively.
― 7 min read
HandFormer improves action recognition using 3D hand poses and images.
― 5 min read
Mamba shows strong potential in video analysis compared to transformers.
― 6 min read
A new method enhances the alignment of 3D body models to 2D images.
― 5 min read
A new text encoder tackles challenges in visual text rendering.
― 4 min read
PosSAM improves image segmentation with open-vocabulary capabilities and innovative techniques.
― 6 min read
A new method enhances the accuracy of names in image segmentation datasets.
― 6 min read
A new method enhances safety features in multimodal AI systems without extensive training.
― 6 min read
SkateFormer enhances human action recognition through advanced joint analysis and attention techniques.
― 6 min read
Exploring how strokes in sketches influence AI decision-making.
― 4 min read
A new Bayesian framework improves image quality and estimates uncertainty.
― 6 min read
A method for assessing the transferability of pre-trained models for object detection.
― 4 min read
A new method separates moving and static elements in city environments.
― 7 min read
RAIN-GS improves 3DGS performance with better initialization and filtering techniques.
― 7 min read
AVIBench tests LVLMs to ensure they withstand adversarial visual instructions.
― 7 min read
SELECTOR utilizes diverse data to predict cancer patient survival accurately.
― 8 min read
Switch-DiT improves image generation through innovative denoising task relationships.
― 6 min read
This article discusses methods to improve deep learning's resilience to adversarial examples.
― 6 min read
Improving robustness against adversarial attacks in vision-language models.
― 4 min read
A new method enhances object detection by addressing domain shifts using attention mechanisms.
― 6 min read
A novel approach improves shape representation in complex 3D data.
― 6 min read
New techniques enhance CBCT image quality and speed for better patient outcomes.
― 6 min read
A new method enhances matching across visible and infrared camera images.
― 5 min read
This study enhances facial AU detection using synthetic data for better accuracy and fairness.
― 5 min read
A new method enhances realistic human images by focusing on hand generation.
― 6 min read
New watermarking method strengthens protection of machine learning models against theft.
― 6 min read
InterLUDE improves model performance using labeled and unlabeled data.
― 6 min read
PALM improves computer vision models' adaptability in changing environments.
― 8 min read
A study on HDMap generation using SDMap prior and its performance evaluation.
― 7 min read