This is the Linux app named DINOv3 whose latest release can be downloaded as dinov3sourcecode.tar.gz. It can be run online in the free hosting provider OnWorks for workstations.
Download and run online this app named DINOv3 with OnWorks for free.
Befolgen Sie diese Anweisungen, um diese App auszuführen:
- 1. Diese Anwendung auf Ihren PC heruntergeladen.
- 2. Geben Sie in unserem Dateimanager https://www.onworks.net/myfiles.php?username=XXXXX den gewünschten Benutzernamen ein.
- 3. Laden Sie diese Anwendung in einem solchen Dateimanager hoch.
- 4. Starten Sie den OnWorks Linux-Online- oder Windows-Online-Emulator oder den MACOS-Online-Emulator von dieser Website.
- 5. Rufen Sie vom gerade gestarteten OnWorks Linux-Betriebssystem aus unseren Dateimanager https://www.onworks.net/myfiles.php?username=XXXXX mit dem gewünschten Benutzernamen auf.
- 6. Laden Sie die Anwendung herunter, installieren Sie sie und führen Sie sie aus.
SCREENSHOTS
Ad
DINOv3
BESCHREIBUNG
DINOv3 is the third-generation iteration of Meta’s self-supervised visual representation learning framework, building upon the ideas from DINO and DINOv2. It continues the paradigm of learning strong image representations without labels using teacher–student distillation, but introduces a simplified and more scalable training recipe that performs well across datasets and architectures. DINOv3 removes the need for complex augmentations or momentum encoders, streamlining the pipeline while maintaining or improving feature quality. The model supports multiple backbone architectures, including Vision Transformers (ViT), and can handle larger image resolutions with improved stability during training. The learned embeddings generalize robustly across tasks like classification, retrieval, and segmentation without fine-tuning, showing state-of-the-art transfer performance among self-supervised models.
Eigenschaften
- Simplified self-supervised learning framework with improved scalability
- Teacher–student distillation without labeled data or heavy augmentation
- Support for multiple backbones including Vision Transformers
- Stable high-resolution training and distributed multi-GPU setup
- High transferability to classification, retrieval, and segmentation tasks
- Ready-to-use scripts for training, feature extraction, and benchmarking
Programmiersprache
Python
Kategorien
This is an application that can also be fetched from https://sourceforge.net/projects/dinov3.mirror/. It has been hosted in OnWorks in order to be run online in an easiest way from one of our free Operative Systems.
