CoolFace
Modelpublic

SharpAI/dinov3-vits16-pretrain-lvd1689m-coreml-FP32-560x560

sourceHugging Faceupdated 11mo agoView on Hugging Face
0likes9downloads
Model Card

DINOv3 VITS16 CoreML FP32

CoreML conversion of facebook/dinov3-vits16-pretrain-lvd1689m optimized for Apple Silicon.

Model Details

  • —Base Model: facebook/dinov3-vits16-pretrain-lvd1689m
  • —Framework: CoreML
  • —Precision: FP32
  • —Input Size: 560×560
  • —Model Size: 83.1 MB

Usage

Python (CoreML)

python
import coremltools as ct
import numpy as np
from PIL import Image

# Load model
model = ct.models.MLModel("dinov3_vits16_560x560_fp32.mlpackage")

# Prepare image
image = Image.open("image.jpg").resize((560, 560))

# Extract features
output = model.predict({"image": image})
features = output["features"]  # Shape: [1, embed_dim, grid_size, grid_size]

Swift/iOS

swift
import CoreML

// Load model
guard let model = try? MLModel(contentsOf: modelURL) else {
    fatalError("Failed to load model")
}

// Prepare image
guard let image = UIImage(named: "image.jpg") else {
    fatalError("Failed to load image")
}

// Extract features
let input = try MLFeatureValue(image: image.cgImage!)
let output = try model.prediction(from: [input])
let features = output.featureValue(for: "features")?.multiArrayValue

Performance

Performance metrics on Apple Silicon:

CoreML Performance

  • —Throughput: 19.02 FPS
  • —Latency: 52.58 ± 0.44 ms
  • —Min Latency: 51.96 ms
  • —Max Latency: 54.40 ms

Speedup vs PyTorch

  • —PyTorch: 9.62 FPS
  • —CoreML: 19.02 FPS
  • —Speedup: 1.98x faster ⚡

Feature Accuracy

  • —Cosine Similarity: 0.9891 (vs PyTorch)
  • —Correlation: 0.9891
  • —Quality: ⭐⭐⭐ Very Good - Excellent similarity

Model Specifications

  • —Precision: FP32
  • —Input Size: 560×560
  • —Model Size: 83.1 MB

License

This model is released under the DINOv3 License. See LICENSE.md for details.

Citation

bibtex
@article{dinov3,
  title={DINOv3: A Versatile Vision Foundation Model},
  author={Meta AI Research},
  journal={arXiv preprint arXiv:2508.10104},
  year={2025}
}

Reference: DINOv3 Paper

Key contributions:

  • —Gram anchoring strategy for high-quality dense feature maps
  • —Self-supervised learning on 1.689B images
  • —Superior performance on dense vision tasks
  • —Versatile across tasks and domains without fine-tuning

Demo Images

Input Image

<div align="center"> <img src="demo_image.png" alt="Demo Input Image" width="500"/> </div> Sample input image for feature extraction demonstration

Feature Visualization

<div align="center"> <img src="dinov3featurecomparison.png" alt="Feature Comparison Visualization" width="800"/> </div>

The visualization shows:

  • —PCA projection of high-dimensional features (RGB visualization)
  • —Feature channel activations showing spatial patterns
  • —Gram matrix analysis for object similarity detection
  • —Side-by-side comparison with PyTorch reference implementation

This comprehensive visualization demonstrates that CoreML conversion preserves the semantic structure and feature quality of the original DINOv3 model.

Powered By DINOv3

🌟 This model is powered by DINOv3 🌟

Converted by Aegis AI for optimized Apple Silicon deployment.

Related Models

Last updated: 2025-10-31