Computer vision and pattern recognition
PulseAugur coverage of Computer vision and pattern recognition — every cluster mentioning Computer vision and pattern recognition across labs, papers, and developer communities, ranked by signal.
- used by Diffusion Models 90%
- instance of Vision--Language Models 90%
- instance of CatalyzeX Code Finder for Papers 70%
- instance of Influence Flower 70%
- instance of arXivLabs 70%
- authored arXivLabs 70%
- instance of Diffusion Models 70%
- instance of RGB color model 70%
- instance of cs.CV 70%
- used by Vision--Language Models 70%
- used by RGB color model 60%
- authored CatalyzeX Code Finder for Papers 50%
12 day(s) with sentiment data
-
AI maps demographic data from 1.3M pathology images
Researchers have developed a method to map demographic information from routine clinical histopathology images. By analyzing 1.3 million whole-slide images from over 250,000 patients across 25 organs, they utilized path…
-
New framework enables video stitching from multiple moving cameras
Researchers have developed MViS, a novel framework for aligning videos captured by multiple moving cameras. This method utilizes a spatial-transformer-inspired network to jointly process frames from all views within a t…
-
Color's role in object recognition under occlusion studied
A new research paper explores how color impacts object recognition in both humans and computer vision systems when objects are partially obscured. The study found that while camouflage occlusions negatively affect human…
-
New AI methods enhance 3D reconstruction from sparse 2D image data
Researchers have developed new methods for improving 3D reconstruction from limited 2D image data, particularly in scenarios where traditional dense scans are not feasible. One approach utilizes pose-aware diffusion pri…
-
Vision models trained on egocentric tennis data predict player and ball movements
Researchers have developed a new method to enable agile robots to perceive and predict dynamic sports actions using egocentric vision. They created a tennis dataset with synchronized video, player positions, and stroke …
-
Survey details methods for accelerating Video Large Language Models
A new survey paper published on arXiv details methods for accelerating inference in Video Large Language Models (VLMs). The paper categorizes techniques based on their intervention points within the inference pipeline, …
-
New unsupervised framework detects retinal anomalies using adapted visual models
Researchers have developed a novel unsupervised framework to detect and localize anomalies in retinal optical coherence tomography (OCT) scans. This approach adapts pretrained visual foundation models to learn normal re…
-
New MERA tensor network method enhances OCT image processing and disease diagnosis
Researchers have developed a novel method using a low-rank multi-scale entanglement renormalization ansatz (MERA) tensor network to process 3D Optical Coherence Tomography (OCT) image volumes. This approach enables join…
-
CompMVR framework uses diffusion models for 3D reconstruction from compressed images
Researchers have introduced CompMVR, a novel framework designed to address the challenges of 3D reconstruction from compressed multi-view images. This method specifically targets the geometric inconsistencies that arise…
-
New PC-DAPS method improves diffusion model sampling for inverse problems
Researchers have introduced Prior-Consistent DAPS (PC-DAPS), a novel diffusion model sampling technique designed to improve the reconstruction of ambiguous nonlinear inverse problems. Unlike existing methods that rely o…
-
MixPT introduces interleaved attention for efficient 3D point cloud processing
Researchers have developed MixPT, a novel point transformer designed to efficiently process 3D data. MixPT combines window attention and global attention mechanisms in an interleaved schedule to balance computational co…
-
New Survey Maps Long-Document Visual Question Answering Landscape
A new survey paper published on arXiv provides a task-centric breakdown of long-document visual question answering (VQA). It categorizes the field into document parsing, evidence organization and retrieval, generator or…
-
New regularization method boosts deepfake detector generalization
Researchers have developed a new method called Hierarchical Spatial Prototype (HSP) regularization to improve the detection of AI-generated face forgeries. This technique addresses the issue of "forensic representation …
-
New method improves AI signal reconstruction by accounting for model uncertainty
Researchers have developed an uncertainty-aware likelihood (UAL) method for diffusion posterior samplers, which are used in signal reconstruction. UAL approximates the marginal likelihood by integrating over uncertain p…
-
New PRAISE framework enhances person re-identification with vision-language models
Researchers have developed a novel framework called PRAISE (Prompt-guided Attribute-based Inference-time Similarity Enhancement) to improve person re-identification (ReID) using vision-language models. This query-side t…
-
New research explores saliency maps for evaluating generative image models
Researchers have explored a novel method for evaluating generative image models by utilizing saliency maps. This approach trains a classifier to distinguish between real images and those generated by a specific model, t…
-
New HiDist Framework Enhances Zero-Shot Anomaly Detection
Researchers have developed a new framework called HiDist for zero-shot anomaly detection, which aims to identify anomalies in unseen categories using auxiliary images. This method leverages distributional prompting and …
-
New HEADA-LTCTPM method improves pedestrian trajectory prediction accuracy
Researchers have developed a new trajectory prediction method called HEADA-LTCTPM to improve accuracy and robustness in pedestrian movement analysis. This method addresses limitations in existing models by considering a…
-
New 3D Reconstruction Model Learns from Motion and Gravity Priors
Researchers have developed a new feed-forward transformer model for 3D reconstruction that incorporates motion and gravity awareness. This model utilizes privileged information, such as optical flow and gravity directio…
-
New FPV-CRASH dataset targets drone crash moment localization
Researchers have introduced FPV-CRASH, a new dataset designed for localizing crash moments in drone first-person-view videos. This dataset addresses a gap in existing resources, which often focus on different types of f…