Fix/kv260 padim xmodel

[DeepKnowledge1]

🔗 Related Issue

Fixes #

📝 Description

• Added KV260 XModel support for PaDiM and PatchCore.
• Added KV260 INT8 quantization and XModel compilation workflows.
• Updated inference backend and documentation with quantization guides.
• Added detailed documentation for running PaDiM and PatchCore on KV260.
• XModel compilation validated successfully with Vitis AI 3.5.
• On-device inference/accuracy validation is pending because KV260 hardware is currently unavailable.

🔄 Type of Change

• 🐛 Bug fix (non-breaking change which fixes an issue)
• 🚀 New feature (non-breaking change which adds functionality)
• 💥 Breaking change
• 📖 Documentation update
• 🏗️ Infrastructure / CI/CD update

🧪 Hardware & Matrix Testing

Vitis AI / KV260:

• PaDiM INT8 calibration
• PaDiM XModel generation
• PaDiM KV260 compilation
• PatchCore XModel generation
• KV260 XModel graph/subgraph inspection
• KV260 on-device inference — hardware unavailable
• ONNX vs XModel accuracy comparison — pending hardware

Host OS used for testing:

• Linux / Ubuntu
• Windows
• macOS

✅ Developer Checklist

• Code follows the project style guidelines.
• uv run pytest — not run in the Vitis AI environment.
• No new Python dependencies added.
• Documentation updated.
• Quantization and compilation workflow manually verified.

📸 Screenshots / Visual Proof

Vitis AI successfully compiled the PaDiM model for:

DPUCZDX8G_ISA1_B4096
KV260

Generated:

PadimKV260.xmodel

On-device validation remains pending due to unavailable KV260 hardware.