A tool that gracefully solves hCaptcha challenges using multimodal large language models, without relying on browser extensions or third-party captcha services.
One-Minute Overview#
hCaptcha Challenger is an innovative captcha solving tool that uses AI technology to automatically complete hCaptcha challenges. The project doesn't rely on browser extensions or third-party captcha services, but instead implements specific interfaces to enable "AI vs AI" solutions. It's suitable for developers, researchers, or automation testers who need to handle web captchas automatically.
Core Value: Through multimodal large language model technology, it provides an automatic captcha solving solution without human intervention.
Core Capabilities#
1. Multiple Captcha Type Support#
- Image Label Binary: Uses ResNet ONNX model for classification
- Image Area Selection (Point Selection): Uses YOLOv8 ONNX detection to identify click areas
- Image Area Selection (Bounding Box): Uses YOLOv8 ONNX segmentation to identify bounding boxes
- Image Drag & Drop: Uses Spatial Chain-of-Thought to implement drag operations
- Multiple Choice: Uses ViT ONNX zero-shot learning for multiple choice questions
2. Advanced Task Processing#
- Rank Strategy: Uses nested model collections to optimize selection strategies
- Self-Supervised Challenge: Uses CLIP-ViT models to handle unknown captcha types
- Agentic Workflow: Uses multimodal large language models for complex task processing
3. Complete Workflow Support#
- Automated CI/CD pipeline: Includes sentinel and collector automation
- Dataset management: Supports VCS and annotation capabilities
- Model training & validation: Provides ResNet and YOLOv8 training/validation workflows