Computer Vision

OneDet3D: A universal 3D object detector trained jointly on diverse indoor/outdoor datasets, achieving one-for-all performance across domains and categories.

Vision Transformers achieve surprisingly high accuracy by transferring only pre-training attention maps, challenging the conventional belief that feature learning is crucial.

MM-Det, a novel algorithm, uses multimodal learning and spatiotemporal attention to detect diffusion-generated videos, achieving state-of-the-art performance on the new DVF dataset.

Researchers achieve state-of-the-art image generation by disentangling semantic and control metadata in diffusion models and optimizing pre-training across resolutions.

ODGS: Lightning-fast 3D scene reconstruction from single omnidirectional images using 3D Gaussian splatting, achieving 100x speedup over NeRF-based methods.

ODGEN: Boosting object detection accuracy by generating high-quality synthetic images using diffusion models conditioned on bounding boxes and text descriptions.

OctreeOcc uses octree queries for efficient and multi-granularity 3D occupancy prediction, surpassing state-of-the-art methods with reduced computational costs.

OccFusion: High-fidelity human rendering from videos, even with occlusions, using 3D Gaussian splatting and 2D diffusion priors.

Neuroscience-inspired motion energy processing enables human-like zero-shot generalization in figure-ground segmentation, outperforming deep learning models on random dot stimuli.

NVRC: A novel end-to-end neural video codec achieves 23% coding gain over VVC VTM by optimizing representation compression.

CompILer: A novel prompt-based incremental learner mastering state-object compositions without forgetting, achieving state-of-the-art performance.

Normal-GS improves 3D Gaussian Splatting by integrating normal vectors into the rendering pipeline, achieving near state-of-the-art visual quality with accurate surface normals in real-time.

Self-supervised gradients boost frozen deep learning model performance!

NeuroGauss4D-PCI masters complex point cloud interpolation using 4D neural fields and Gaussian deformation fields, achieving superior accuracy in dynamic scenes.

NeuRodin: A two-stage neural framework achieves high-fidelity 3D surface reconstruction from posed RGB images by innovatively addressing limitations in SDF-based methods, resulting in superior reconst…

Neural SDF inference is revolutionized by dynamically aligning 3D Gaussians to a neural SDF’s zero-level set, enabling accurate, smooth 3D surface reconstruction.

Neural-RDM: A novel framework for deep, scalable vision generation using residual diffusion models, achieving state-of-the-art results on image and video benchmarks.

Learn disentangled 3D object poses and transfer them between different object identities using a novel neural pose representation, boosting 3D shape generation!

Neural Localizer Fields (NLF) revolutionizes 3D human pose and shape estimation by learning a continuous field of point localizer functions, enabling flexible training on diverse data and on-the-fly p…

Neural Isometries learns a latent space where geometric relationships in the observation space are represented as isometries in the latent space, enabling efficient handling of complex symmetries and …