arXiv Computer Vision By Linman Wang, ZiFei Zhang, Chunran Zheng, Xiwang Dong, Jiarong Lin

DIFTA-3D: Depth-Consistent Instance-Level Feature Transfer and Adaptation of DINOv3 for 3D Detection

Read the original on arXiv Computer Vision →

The paper introduces DIFTA-3D, a method that replaces the task‑specific visual branch in IIFNet3D with a frozen DINOv3 foundation model for RGB‑D 3D instance detection. It employs a depth‑consistent feature pipeline that projects points into calibrated RGB‑D frames, filters features with a metric depth‑residual check, caches accepted DINOv3 features, and aggregates them within proposal‑aligned RoI grids. Extensive experiments on ScanNetV2 show that the DINOv3 control achieves mAP scores of 76.15/60.93 at IoU thresholds 0.25/0.50, while the Conservative VAID recipe improves these to 76.59/62.16, indicating a modest gain from the proposed transfer recipe.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Computer Vision.

arXiv Machine Learning
Oct 9

One Block, Multiple Depths: Recurrent Vision Transformers with Depth-Programmed Experts

The paper introduces reViT, a recurrent vision transformer that uses a single Transformer block repeatedly to achieve the accuracy of a full-depth encoder while keeping inference FLOPs comparable. It achieves depth‑specific transformations by expressing each block’s feed‑forward network as a convex combination of a small shared expert bank, with a continuous depth coordinate guiding the mixture. Experiments on ImageNet‑1k and DINOv2 distillation show that reViT matches or surpasses state‑of‑the‑art models, reduces stored parameters by about 70%, and supports elastic‑depth deployment from a single checkpoint.

By Adrian Bulat, Yassine Ouali, Georgios Tzimiropoulos
Hugging Face Trending Papers
4d ago

HiVE-VQA: Hierarchical Interaction via Volumetric Embedding for Video Quality Assessment

HiVE-VQA introduces a new approach to no-reference video quality assessment by leveraging a single frozen JEPA-based video foundation model. The method extracts token grids from multiple depths of a pre‑trained V-JEPA 2.1 encoder, stacks them into a hierarchical spatio‑temporal volume, and fuses them with a lightweight 3D convolutional head. With only 5.2 M trainable parameters, HiVE-VQA outperforms 17 state‑of‑the‑art methods on both intra‑dataset and cross‑dataset benchmarks, demonstrating that depth‑wise hierarchical fusion from a frozen backbone is both effective and efficient.