AI_WQ | Computer Vision & Generative AI

Domain Overview

AI_WQ is a Project Phoenix domain for WorldQuant University's AI Lab curriculum covering deep learning for computer vision. It implements the complete Four-Phase Agentic Framework with CLI, Cockpit GUI, and unified agent architecture.

The domain spans 6 projects from basic image classification through generative models, providing a comprehensive exploration of modern computer vision techniques.

Project Catalog

Proj1: Image Classification

CNN from scratch for wildlife camera trap images. Binary and multiclass classification.

CNN Binary Multiclass

View Details

Proj2: Transfer Learning

Fine-tuning pretrained models for cassava disease detection with callbacks.

Fine-tuning Pretrained Callbacks

View Details

Proj3: Object Detection

YOLOv8 for object detection with data augmentation techniques.

YOLOv8 Detection Augmentation

View Details

Proj4: Face Recognition

MTCNN and InceptionResNet for face detection with Flask API deployment.

MTCNN InceptionResNet Flask API

View Details

Proj5: GANs

Generative Adversarial Networks with generator and discriminator training.

Vanilla GAN Generator Discriminator

View Details

Proj6: Diffusion Models

Stable Diffusion for text-to-image generation with prompt engineering.

Stable Diffusion Prompting Text-to-Image

View Details

Quick Start

CLI (Phase 0)

cd domains/AI_WQ/cli python main.py AI_WQ> list projects AI_WQ> info 1 AI_WQ> models AI_WQ> suggest detection AI_WQ> compare 1 3

Cockpit GUI (Phase 1-4)

cd domains/AI_WQ/cockpit python ai_wq_cockpit.py

Features: Blueprint Panel, Execution Trace, Inspector Panel, Artifact Workspace, Pause/Resume/Cancel

Agentic Interface

from unified_agent import AIDataClient, ToolRegistry, ExecutionContext from unified_agent import PlanBuilder, AgenticEngine # Initialize client = AIDataClient() context = ExecutionContext() tools = ToolRegistry(client, context) engine = AgenticEngine(tools, context) # Parse query and execute plan = engine.parse_query("explore project 1 image classification") engine.approve_plan() result = engine.execute()

Key Features

Cross-Project Analysis

Compare techniques across all 6 projects

Model Discovery

Find pretrained models and weights

Technique Suggestions

Get recommendations for CV problems

Glass Box Execution

Preview plans before running