requirements.txt

# ===============================
# Core Dependencies
# ===============================
# Computer vision & image processing
opencv-python>=4.8.0
numpy>=1.24.0
scipy>=1.11.0
scikit-image>=0.21.0
scikit-learn>=1.3.0
Pillow>=10.0.0
matplotlib>=3.7.0

# OCR
pytesseract>=0.3.10

# ===============================
# Dimension & Feature Analysis
# ===============================
# Torch is used ONLY for:
# - Device detection (CPU / CUDA)
# - Future optional learning-based extensions
# Current MVP DOES NOT require training or inference with Torch.
torch>=2.0.0
torchvision>=0.15.0

# ===============================
# LLM Integration (Optional)
# ===============================
# OpenAI is used as a reference implementation for:
# - Multimodal (image + structured vector) reasoning
# - Converting perception outputs into language-based descriptions
#
# IMPORTANT:
# - OpenAI is NOT a hard dependency of the architecture.
# - The VisionPipeline is intentionally LLM-agnostic.
# - Any LLM provider (Anthropic, Gemini, local VLMs, etc.)
#   can be integrated by replacing the LLMIntegration module.
#
# If you do not use LLM features, this dependency is OPTIONAL.
openai>=1.0.0

# ===============================
# Optional / Future Extensions
# ===============================
# These are NOT required for the MVP.
# They are listed here to document possible research directions.
#
# transformers>=4.30.0       # For local or open-source multimodal models
# depth-anything-v2>=0.1.0   # For explicit depth estimation (future 3D extensions)