AI-Projects: A Full-Stack AI Project Collection Covering CV, NLP, and LLM

AI-Projects is a comprehensive AI project repository maintained by developer Fawaz Allan, covering fields like computer vision (CV), natural language processing (NLP), and large language models (LLM). It includes practical projects such as intelligent traffic signal control, Discord Gemini bot, and image caption generation, providing a reference sample library and project inspiration for AI learners and developers.

This repository collects practical projects across multiple domains, from traditional RNN to the latest LLM applications, demonstrating the implementation capabilities of AI technology in different scenarios. It is suitable for developers learning AI development or seeking project inspiration.

Intelligent Traffic Signal Control System

• Tech stack: YOLOv11 (object detection), OpenCV (image processing), Gradio (interactive interface), license plate OCR
• Logic: Real-time traffic flow detection, dynamically adjust signal light duration

Discord Gemini 2.0 Bot

• Tech: Google Gemini 2.0 Flash model, Discord.py framework
• Features: Multimodal input (text/image/PDF), OCR extraction, context awareness

BLIP-2 Image Caption Generator

• Architecture: BLIP-2 (Q-Former bridges vision and LLM)
• Implementation: PyTorch + Transformers, Beam Search decoding

BlenderBot Chatbot

• Architecture: Flask backend API + Web frontend + BlenderBot model
• Value: Entry-level template for full-stack AI application development

• Intelligent traffic system: Theoretically reduces average waiting time during peak hours; license plate OCR supports violation tracking/parking lot management expansion
• Discord bot: Achieves deep integration of large language models with instant messaging platforms, supporting multi-turn coherent communication
• BLIP-2 generator: Suitable for image alt text generation, image search, and visual impairment assistance scenarios
• BlenderBot: Provides a complete example of front-end and back-end integration, facilitating full-stack AI development

1. Multimodal capability becomes standard: Gemini bot and BLIP-2 reflect the trend of text+vision integration
2. Collaboration between large and small models: Cloud-based large models (Gemini) combined with local lightweight models (YOLOv11) handle complex reasoning and real-time tasks respectively
3. Engineering delivery: Projects move from experiment to productization through Gradio interfaces and actual platform deployment

Target Audience:

1. CV field: Learn the implementation process of YOLOv11
2. NLP/conversation system field: Understand cloud API calls and local model deployment
3. Full-stack developers: Refer to BlenderBot's front-end and back-end integration

Recommended Learning Order:

1. BLIP-2 image captioning (independent code, easy to understand)
2. Discord bot (API integration and asynchronous processing)
3. Intelligent traffic project (complete CV engineering pipeline)

• Traffic project: Multi-intersection collaborative control, real-time map data access, predictive signal scheduling
• Discord bot: Voice conversation, tool calls (code execution/database query)
• Image captioning: Image search engine, social media tag generation, content moderation assistance

The AI-Projects repository covers multiple typical scenarios of AI application development. Each project focuses on a specific problem domain, providing complete ideas from model selection to engineering implementation. It helps developers understand the trade-offs of technical solutions and apply research models to real user scenarios, serving as a starting point for transforming AI capabilities into practical products.