🌐 core-dino | Resolution-Agnostic Self-Supervised Learning on Satellite Imagery

🔭 Overview

🌀 Quickstart

🧠 Architecture: DiNO × YOLO × I-JEPA

1️⃣ Multi-Resolution DINO Setup (instead of local-global)

In standard DINO / DINOv2, the student sees cropped or distorted views (local), while the teacher sees global views.
In core-dino, we replace this with clean vs degraded resolution contrast:

• 👨‍🏫 Teacher gets clean 30 cm satellite imagery.
• 👨‍🎓 Student sees augmented versions of the same scene at varying resolutions (30 cm → 2 m) with photometric and spatial distortions.

2️⃣ I-JEPA-Style Patch Dropping

• Random patch regions are dropped from the student input.
• The objective is to align the visible patch embeddings with the teacher’s corresponding high-resolution ones.
• This enforces local-global and partial-whole consistency in the latent space.

3️⃣ YOLOv11-X as Encoder Backbone

• We use YOLOv11-X, one of the most powerful and recent YOLO variants, as the spatial encoder.
• The backbone is truncated after 23 layers, retaining rich spatial semantics while maintaining efficiency.
• This provides strong priors from supervised detection tasks, now adapted for self-supervised learning.

🧪 Training Flow: Resolution-Agnostic DiNO

👨‍🏫 1. Teacher View (Clean & High-Res)

• Receives a clean 30 cm image without any augmentation.
• Used as the stable reference to guide the student.

👨‍🎓 2. Student View (Augmented Multi-Resolution)

• Receives randomly augmented versions of the same image:
  • Downsampled to 30 cm to 2 m
  • Augmented with noise, blur, color jitter, spatial dropout, etc.
• Emulates resolution variability common in EO imagery.

⚠️ 3. Spatial Misalignment & Solution

• Since different student resolutions produce different spatial dimensions (H × W),
  we use bilinear interpolation to resize the student’s feature map to match the teacher's spatial shape before computing the contrastive loss.

🎯 4. Objective

• Align the spatial token embeddings of the student with the teacher — pixel-to-pixel and semantically — despite resolution gaps and augmentations.
• Encourages scale-invariant, robust feature learning across real-world variations.

📈 Performance: Latent Quality & Downstream Evaluation

🛣️ Downstream: Road Extraction (DeepGlobe Dataset)

• Setup:
  • Both core-dino and YOLOv11-X backbones were frozen
  • Only a 2-layer convolutional head was trained
  • Task: Binary road segmentation using IoU loss
• Result:
  • core-dino consistently outperformed the supervised YOLOv11-X backbone across all epochs
  • Shows superior latent representation quality, even without task-specific supervision
  • Demonstrates better generalization and semantic robustness in downstream transfer tasks

📓 Reproduce this comparison in Colab:

🏙️ Downstream : Building Footprint Validation

• Setup:
  • Compared YOLOv11-X (original weights) vs. YOLOv11-X initialized with core-dino weights
  • Used same training pipeline for both
• Result:
  • core-dino achieved +15 mAP improvement over standard YOLOv11-X
  • Captures edge-localized and compact building structures better
  • Demonstrates strong spatial precision and high-quality feature encoding

📓 Reproduce this comparison in Colab:

🗂️ Model Details

💳 License

✅ Free to use, share, and adapt for non-commercial research
❌ Commercial use is not permitted without explicit permission
📌 Please provide appropriate credit when using this dataset in publications or projects.