KoharuLayout-RFDETR-Seg-2XL (ONNX Runtime Export)
This repository hosts the compiled ONNX Runtime (.onnx) export of KoharuLayout-RFDETR-Seg-2XL-1152 for lightweight, zero-Python embedded inference in native applications (such as XianScan).
📜 Credits & Acknowledgments
All credit for the model architecture, training, and supervision belongs to:
- Model Creator: mayocream
- Original Model Weights (SafeTensors): mayocream/koharu-layout-rfdetr-seg-2xl-1152
- Original Project: Koharu (Manga Translation Studio)
- Dataset: Manga109 Segmentation v2.0.0 (Manga109 academic-use terms apply)
We express our deepest gratitude to mayocream for their groundbreaking open-source contributions to comic AI layout analysis.
🎯 Target Classes
The model predicts bounding boxes and instance segmentation masks across 4 classes at 1152x1152 resolution:
| ID | Class | Description |
|---|---|---|
| 0 | text |
Dialogue, captions, titles, credits, and general text |
| 1 | onomatopoeia |
Comic sound effects and stylized text (COO / SFX) |
| 2 | bubble |
Speech and dialogue balloons |
| 3 | panel |
Manga panel frames and layout boundaries |
🛠️ Export Details
- File:
rfdetr-seg-2xlarge.onnx(~148 MB) - Input:
images[1, 3, 1152, 1152]float32 (RGB, normalized 0 to 1),orig_target_sizes[1, 2]int64 - Runtime: ONNX Runtime (CPU, DirectML, CUDA, CoreML)