Model Detail
dinov3-vits16-pretrain-lvd1689m
—dinov3-vits16-pretrain-lvd1689m is an AI model with 7B parameters released by Meta. The model is registered under the image-feature-extraction pipeline tag on Hugging Face, distributed under a other license.
dinov3-vits16-pretrain-lvd1689m ships with 7B parameters. Access is gated on Hugging Face under the other license, which means a manual approval step before weights can be downloaded.
Downloads of dinov3-vits16-pretrain-lvd1689m have moved -32.9% over the trailing thirty days. That is a slight downtrend, consistent with normal cooling as newer models compete for the same workloads. These numbers are signal, not guarantee — week-over-week download counts on Hugging Face also reflect mirror traffic, CI scrapes, and one-off benchmarking runs.
dinov3-vits16-pretrain-lvd1689m is best fit for workloads that match the image-feature-extraction pipeline tag, and semantic search, retrieval, and clustering pipelines. Treat this as a starting matrix rather than a benchmark verdict — the right deployment usually depends on the specific evaluation suite that mirrors your workload.
Lightweight Distillation of SAM 3 and DINOv3 for Edge-Deployable Individual-Level Livestock Monitoring and Longitudinal Visual Analytics
arXiv:2604.27128v2 Announce Type: replace-cross Abstract: Foundation-model pipelines for individual-level livestock monitoring -- combining open-vocabulary detection, promptable video segmentation, and self-supervised visual embeddings -- have raised the accuracy ceiling of precision livestock farmi
Synergistic Foundation Models for Semi-Supervised Fetal Cardiac Ultrasound Analysis: SAM-Med2D Boundary Refinement and DINOv3 Semantic Enhancement
arXiv:2605.19799v1 Announce Type: cross Abstract: We present a semi-supervised framework for joint segmentation and classification of fetal cardiac ultrasound images. Built upon the EchoCare multi-task backbone, our method integrates SAM-Med2D for boundary refinement and leverages DINOv3 to enhance
DINORANKCLIP: DINOv3 Distillation and Injection for Vision-Language Pretraining with High-Order Ranking Consistency
arXiv:2605.06592v1 Announce Type: cross Abstract: Contrastive language-image pretraining (CLIP) suffers from two structural weaknesses: the symmetric InfoNCE loss discards the relative ordering among unmatched in-batch pairs, and global pooling collapses the visual representation into a semantic bot
Resolution scaling governs DINOv3 transfer performance in chest radiograph classification
arXiv:2510.07191v3 Announce Type: replace-cross Abstract: Self-supervised learning (SSL) has improved visual representation learning, but its value in chest radiography remains uncertain. DINOv3 extends earlier SSL models through Gram-anchored self-distillation and explicit high-resolution adaptatio
