OpenCV MCP Server
by GongRzhe
Image and video processing with computer vision through AI assistants
ai-ml Python Intermediate Self-hostable No API key
⭐ 96 stars 📅 Updated: 6mo ago
Description
An MCP server providing OpenCV's image and video processing capabilities. Offers 22+ tools spanning image basics (resize, crop, color conversion), processing (filters, edge detection, thresholding, contours), computer vision (feature detection with SIFT/ORB/BRISK, face detection, YOLO object detection), and video processing (frame extraction, motion detection, object tracking, webcam integration). Supports real-time camera detection and video output generation.
✅ Best for
Developers needing computer vision capabilities accessible through AI assistants
⏭️ Skip if
You only need basic image resizing or prefer a cloud-based vision API
💡 Use cases
- Analyzing and processing images through AI assistant commands
- Detecting objects and faces in images and video streams
- Extracting and matching features across multiple images
- Processing video for motion detection and object tracking
- Real-time webcam-based object detection
👍 Pros
- ✓ Comprehensive 22+ tool set covering image, video, and computer vision
- ✓ Real-time camera integration for live processing
- ✓ Multiple detection methods (SIFT, ORB, BRISK, YOLO, Haar cascades)
- ✓ No API key required — fully local processing
👎 Cons
- ✗ Heavy dependencies (OpenCV, numpy, optional DNN models)
- ✗ YOLO model files need separate download for object detection
- ✗ Video processing can be CPU-intensive
🔧 Exposed tools (11 tools)
| Tool | Category | Description |
|---|---|---|
| detect_features_tool | computer-vision | Feature extraction using SIFT, ORB, and BRISK algorithms |
| detect_faces_tool | computer-vision | Face detection via Haar cascades and DNN models |
| detect_objects_tool | computer-vision | YOLO-based object recognition in images |
| resize_image_tool | image-basics | Scale images to specified dimensions |
| apply_filter_tool | image-processing | Apply blur, gaussian, median, and bilateral filters |
| detect_edges_tool | image-processing | Edge detection via Canny, Sobel, Laplacian, and Scharr methods |
| detect_contours_tool | image-processing | Locate and visualize contours in images |
| extract_video_frames_tool | video-processing | Extract frame sequences from video files |
| detect_motion_tool | video-processing | Analyze motion between consecutive video frames |
| track_object_tool | video-processing | Multi-frame object trajectory monitoring |
| detect_camera_objects_tool | video-processing | Real-time webcam object detection |
💡 Tips & tricks
Set OPENCV_DNN_MODELS_DIR to your models directory for YOLO detection. Use detect_features_tool with SIFT for best accuracy or ORB for speed.
Quick info
- Author
- GongRzhe
- License
- MIT
- Runtime
- Python 3.10+
- Transport
- stdio
- Category
- ai-ml
- Difficulty
- Intermediate
- Self-hostable
- ✅
- Auth
- —
- Docker
- —
- Version
- 1.0.0
- Updated
- Sep 11, 2025
Client compatibility
- ❓ Claude Code
- ❓ Cursor
- ❓ VS Code Copilot
- ❓ Gemini CLI
- ❓ Windsurf
- ❓ Cline
- ❓ JetBrains AI
- ❓ Warp
Platforms
🍎 macOS 🐧 Linux 🪟 Windows