# Face++ / Megvii MCP > Face++ / Megvii provides advanced computer vision tools for analyzing images and videos. Your agent can detect faces, compare identities with high confidence scores, analyze human body skeletons, or identify specific hand gestures. It's built for security compliance, identity verification, and detailed biometric research. ## Overview - **Category:** image-video - **Price:** Free - **Tags:** computer-vision, facial-recognition, identity-verification, biometrics, image-analysis, gesture-detection ## Description Your agent connects to Face++ / Megvii to handle complex image analysis that used to require specialized software and manual workarounds. Instead of navigating a dense web console, you talk to your AI client and ask it to perform vision tasks—whether checking KYC documents or analyzing user behavior in a video feed. The system instantly detects faces, calculates how similar two people are, and even maps out human body skeletons for posture analysis. This level of deep visual intelligence is now accessible through the Vinkius catalog, letting you treat complex biometrics like any other data query. You can manage massive face databases, running searches across thousands of stored profiles with a simple command. ## Tools ### add_face_to_faceset Adds a new face to an existing searchable collection of faces (FaceSet). ### compare_faces Compares two specific faces and returns a score indicating how likely they are to be the same person. ### create_faceset Creates a brand new, empty searchable database for storing face profiles. ### detect_body Scans an image and returns the coordinates for any detected human bodies. ### detect_face Locates all visible faces within an image, providing basic attributes like age or gender. ### gesture_detect Analyzes hand positions in an image to identify specific gestures. ### get_faceset_detail Retrieves comprehensive details and metadata for a specified FaceSet. ### remove_face_from_faceset Deletes a specific face profile from an existing collection (FaceSet). ### search_face Searches for a given face within a FaceSet and returns matching individuals. ### skeleton_detect Maps out the full human skeleton keypoints from an image to analyze posture and movement. ## Prompt Examples **Prompt:** ``` Detect faces in this image URL: [URL]. ``` **Response:** ``` I've analyzed the image. I found 2 faces. Face 1 appears to be a male around 30 years old with a happy expression. Face 2 is a female around 28 years old with a neutral expression. Would you like to see the face tokens for these results? ``` **Prompt:** ``` Compare these two images to see if they are the same person: [URL1] and [URL2]. ``` **Response:** ``` I've compared the faces. There is a 98.5% confidence that both images belong to the same person. This is well above the recommended threshold for matching identities. ``` **Prompt:** ``` Check for any human body detected in this photo: [URL]. ``` **Response:** ``` I've retrieved the body detection results. I found 1 human body in the image. Would you like to analyze the skeleton keypoints or check for any specific hand gestures? ``` ## Capabilities ### Identify and detail faces Detects human faces in an image and retrieves attributes like age, gender, and emotional state. ### Verify identity similarity Compares two separate images to calculate the mathematical confidence that they belong to the same person. ### Search large face databases Creates and manages searchable collections of faces, allowing you to look up specific individuals within a group. ### Analyze body structure Detects human bodies in an image and maps out the complete skeleton keypoints for posture analysis. ### Recognize hand movements Identifies specific gestures and patterns within hands captured in the images. ## Use Cases ### Onboarding new employees for compliance A Compliance Officer needs to verify a batch of documents. They tell their agent: 'Compare these 20 IDs using compare_faces and report any matches against our master FaceSet.' The agent runs the checks instantly, flagging discrepancies where human eyes would struggle with volume. ### Analyzing user interactions in product testing A UX Researcher reviews video footage of a usability test. They ask their agent to detect_body and skeleton_detect on key moments. This reveals if the user is hunched over or using specific hand gestures, data that informs immediate design changes. ### Forensic investigation of suspects A Security Analyst receives a photo of an unknown individual. They use detect_face to pull key attributes (age, gender) and then execute search_face against a known database using the FaceSet tools to narrow down suspects. ### Monitoring remote workers for safety A manager needs to check compliance on body posture in factory video feeds. They instruct their agent to detect human bodies and run skeleton_detect, immediately spotting anyone whose posture deviates from safe guidelines. ## Benefits - Instant identity confirmation. Instead of manual visual checks, use compare_faces to get a high-confidence percentage score that two people match. - Deep database management. Use create_faceset and add_face_to_faceset to build structured profile libraries, then search for matches using search_face. - Complete body analysis. Detect human bodies with detect_body, then dive deeper by running skeleton_detect to analyze posture or gesture_detect to check hand signals. - Faster compliance checks. Automate KYC workflows. Your agent can use detect_face immediately on uploaded IDs to confirm basic attributes like gender and age. - Unified workflow. You don't have to switch between multiple image processing tools; your agent handles face, body, and gesture analysis all at once. ## How It Works The bottom line is that you get deep computer vision analysis directly through natural language prompts from your AI client. 1. Subscribe to this MCP, then enter your Face++ API Key and Secret into your preferred AI client. 2. Ask your agent to perform a specific vision task—for example, comparing two photos or detecting bodies in an image URL. 3. The tool processes the data and returns structured results, such as confidence scores or detected attributes. ## Frequently Asked Questions **How does Face++ / Megvii MCP handle face comparisons?** It compares two faces and outputs a confidence percentage. The compare_faces tool gives you a score, not just a yes or no answer, helping determine if the match is reliable. **Can I use Face++ / Megvii MCP to track people over time?** You can manage persistent identity records using create_faceset and add_face_to_faceset. This allows your agent to build a searchable history of profiles for longitudinal analysis. **Do I need to write code to analyze body posture?** No, you don't. You just tell your agent: 'Detect the body in this image and run skeleton_detect.' The MCP handles the complex steps of mapping keypoints for you. **What is the difference between detect_face and search_face?** Detect_face simply finds all faces in a single picture, giving basic attributes. Search_face requires you to first build a FaceSet and then searches that existing library for matches. **Which tool is best for checking hands or gestures?** Use gesture_detect if you are looking for specific hand signs (like counting or pointing). It focuses solely on identifying those recognized movements from the image data.