Imagerecog

🚀 BenBox

A proxy AI system for an MCP tool based on SSE, including a mobile Angular application.

Why Choose MCP?

The MCP server can now be a running process, and clients can connect to it, use it, and disconnect at any time. In other words, the SSE-based server and client can be decoupled processes (possibly even on different nodes). This is different from the STDIO-based mode, where the client itself starts the server as a subprocess. This approach is more suitable for "cloud-native" use cases.

🚀 Quick Start

📦 Installation

Install the required packages and download the MCP server and client:

# Install Node.js on Linux
sudo apt install nodejs npm

# Install Node.js on Mac
brew install nodejs npm

# Install mcp
conda install -c conda-forge mcp 

# Clone the repository
git clone https://github.com/DrBenjamin/imagerecog

💻 Usage Examples

Basic Usage

Test image bytes for testing in MCP Inspector or VS Code Copilot Chat:

"iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAQAAAC1HAwCAAAAC0lEQVR4nGNgYAAAAAMAAWgmWQ0AAAAASUVORK5CYII="

Test image URLs:

https://www.seriousbenentertainment.org/R/Image.png
http://212.227.102.172:9000/templategenerator/subfolder/City.Golf.jpg

Advanced Usage

Run the MCP server and Streamlit client application:

# Create a conda environment using the environment.yml file
conda env create -f environment.yml

# Activate the conda environment
conda activate imagerecog
python -m pip install "mcp[cli]"

# 1. Run the MCP development server
mcp dev src/server.py

# 2. Run the MCP server
python src/server.py

# 3. Run the Streamlit application
python -m streamlit run mobile/app.py

# Or run the script
sudo chmod 755 run.sh
./run.sh
lsof -i :6274
lsof -i :8080
lsof -i :8501

⚙️ Configuration

Change the configuration and add your OpenAI API key in the .streamlit/st.secrets.toml file:

# LLM provider
LLM_LOCAL = "False"  # `False` means using the local Ollama model, `True` means using the OpenAI API

# MCP API
[MCP]
MCP_URL = "http://127.0.0.1:8080"
MCP_SYSTEM_PROMPT = "<System prompt for image recognition>"
MCP_USER_PROMPT = "<User prompt for image recognition>"

# Ollama API
[OLLAMA]
OLLAMA_URL = "http://127.0.0.1:11434"
OLLAMA_MODEL = "<ollama model>" # e.g., llava or "llama3.2-vision"

# OpenAI API
[OPENAI]
OPENAI_API_KEY = "sk-..."
OPENAI_MODEL = "<model>" # e.g., "o4-mini" or "gpt-4.1" or "gpt-4o" or "gpt-4-turbo"

🤖 Ollama

To install and run the Ollama model, use the following commands:

# Run the Ollama server
conda install -n ollama -y --strict-channel-priority python=3.8.10
git clone https://github.com/ollamaai/ollama.git
cd ollama/
pip install -r requirements.txt
python server.py

# Install the model
ollama pull mistral-7b-instruct

📱 Mobile Build

To build the application for mobile devices, use the following commands (make sure the Capacitor version matches):

# Initialize the Capacitor project in the project root directory
cd mobile
npm install @capacitor/core@5.7.8 @capacitor/cli@5.7.8
npm install @capacitor/ios@5.7.8
npm install @capacitor/browser@^5 --save
npx cap init BenBox org.benbox.benbox --webDir=../browser
npx cap add ios
# Or
npx cap sync ios
cd ios/App
pod install

# Build the application (from the project root directory)
npm install --prefix mobile
npm run build --prefix mobile
npm run cap:copy --prefix mobile
npm run cap:open:ios --prefix mobile