Build Lobster AI System from Scratch: Beginner-Friendly Tutorial for 5 Major Companies' Coding Plans

2026-03-16

Hello everyone, this is programmer Wan Feng actively working on various AI projects.

In my previous article I compared 5 major companies' Lobster Coding Plans, many readers asked: "Wan Feng, stop talking and show us a practical tutorial!"

Okay, today's beginner-friendly tutorial takes you from scratch, using 5 companies' tools to build a lobster freshness detection system.

Tutorial Features:

✅ Every step has code and screenshots
✅ Provides complete runnable code
✅ Contains common errors and solutions
✅ Suitable for zero-basis beginners

📋 Preparation Work

Environment Requirements

Item	Requirement
Python	3.9+
Memory	8GB+
Hard Drive	20GB+
Network	Can access internet

Register Accounts

Need to register the following platform accounts in advance:

Prepare Dataset

I compiled a public lobster dataset (1000 images, labeled):

Download: https://github.com/wanfeng-ai/lobster-dataset
Contains: A/B/C/D four grades, fresh/not fresh two categories

🦞 Solution 1: Tencent OpenClaw Implementation (Recommended)

Step 1: Install OpenClaw

# Install OpenClaw CLI
pip install openclaw

# Initialize project
openclaw init lobster-detector
cd lobster-detector

Step 2: Configure Model

# Edit configuration file
cat > config.yaml << EOF
model:
  provider: tencent
  name: hunyuan-vision-pro
  api_key: your_Tencent_Cloud_API_key

dataset:
  path: ./data/lobster
  train_split: 0.8

task:
  type: image_classification
  labels: ["fresh", "not_fresh", "grade_A", "grade_B", "grade_C", "grade_D"]
EOF

Step 3: Write Skill

# skills/lobster_detector.py
from openclaw import skill
import requests

@skill.register
def detect_lobster_freshness(image_url: str) -> dict:
    """
    Detect lobster freshness

    Args:
        image_url: Lobster image URL

    Returns:
        Dictionary containing freshness and grade
    """
    # Call Tencent Hunyuan Vision API
    response = requests.post(
        "https://api.tencent.com/vision/lobster",
        json={
            "image_url": image_url,
            "task": "freshness_detection"
        },
        headers={
            "Authorization": f"Bearer {YOUR_API_KEY}"
        }
    )

    result = response.json()

    return {
        "fresh": result["confidence"] > 0.85,
        "grade": result["grade"],  # A/B/C/D
        "confidence": result["confidence"],
        "suggested_price": result["price_suggestion"]
    }

jsonContent: meta: false pages: false posts: title: true date: true path: true text: false raw: false content: false slug: false updated: false comments: false link: false permalink: false excerpt: false categories: false tags: true