DenseMarks

A PyTorch implementation for dense UVW coordinate prediction from human head images using a DINOv3 backbone and a DPT-style head architecture.

Overview

DenseMarks predicts per-pixel positions in the canonical space (cube $[0, 1] ^ 3$) from human head images.

• Input: RGB image of size 512×512
• Output: UVW coordinate tensor (B, 3, 512, 512) with values in [0, 1]

---

🚀 Current Status

DenseMarks currently supports inference only — you can run the model to generate dense UVW predictions from input images.
🧠 Training support is coming soon! Stay tuned :)

---

Prerequisites

• Python 3.8+
• PyTorch 1.12+
• CUDA (optional, for GPU acceleration)

Installation

1. Clone the repository:

   git clone https://github.com/diddone/densemarks.git
   cd densemarks

2. Install DINOv3 submodule:

   git clone https://github.com/facebookresearch/dinov3 third_party_dinov3

3. Modify DINOv3 for compatibility:

   # For Linux (GNU sed):
   sed -i '/dinov3\.hub\.segmentors/s/^/#/; /dinov3\.hub\.classifiers/s/^/#/; /dinov3\.hub\.detectors/s/^/#/; /dinov3\.hub\.dinotxt/s/^/#/; /dinov3\.hub\.depthers/s/^/#/' third_party_dinov3/hubconf.py
   
   # For macOS (BSD sed):
   sed -i '' '/dinov3\.hub\.segmentors/s/^/#/; /dinov3\.hub\.classifiers/s/^/#/; /dinov3\.hub\.detectors/s/^/#/; /dinov3\.hub\.dinotxt/s/^/#/; /dinov3\.hub\.depthers/s/^/#/' third_party_dinov3/hubconf.py

4. Install dependencies:

   pip install torch transformers numpy

5. Download model weights from Hugging Face:

   from dense_marks_model import DenseMarksModel, read_image
   from huggingface_hub import hf_hub_download
   model = DenseMarksModel(hf_hub_download("diddone/densemarks", "model.safetensors"))
   images = read_image("assets/00000.png") # rgb, 512x512
   uvw = model(images) # Predict UVW coordinates