🛡️ Revpass-Single (YOLOv26s-cls)

Revpass-Single is a highly optimized single-tile classifier designed to identify reCAPTCHA v2 tile contents. It is a core component of the Revpass AI solver system.

📊 Performance

• Model Architecture: YOLOv26s-cls (Medium)
• Top-2 Accuracy: 90.0% (Verified on Stratified Validation Set)
• Use Case: Filtering "Best Match" tiles for 4x4 grids.

🚀 Usage

from ultralytics import YOLO
# Load the model
model = YOLO("[https://huggingface.co/saifyxpro/revpass-single/resolve/main/revpass-single.pt](https://huggingface.co/saifyxpro/revpass-single/resolve/main/revpass-single.pt)")
# Inference
results = model("path/to/tile.jpg")
print(results[0].probs.top1conf)
📂 Files
revpass-single.pt: PyTorch weights (Best).
revpass-single.onnx: ONNX export for high-performance inference.
Generated by Revpass Auto-Trainer