Tencent Cloud Face Detection
Paste the following prompt into your AI chat to install this skill:
Please install @tencent-adm/tencentcloud-faceid-detectface according to https://skillhub.cn/install/skillhub.md.
About this skill
Problem to Solve
When an image enters a business pipeline, the first checks are often: are there faces, how many, where are their bounding boxes, and are they clear and frontal enough. Plain image processing rarely provides attributes, pose, and quality signals together, while direct face recognition may add unnecessary identity-matching cost. tencentcloud-faceid-detectface targets these detect-first use cases by wrapping Tencent Cloud IAI DetectFace as a scriptable skill.
How It Works
The skill accepts a local image path, Base64, or image URL, chooses the input according to priority, and can return multiple result sets:
- Face localization: returns X, Y, Width, and Height, useful for cropping or annotation.
- Multi-face detection: --max-face-num controls the maximum number of faces, default 1 and maximum 120.
- Face attributes: enabling --need-face-attributes returns gender, age, expression, beauty, glasses, mask, Pitch, Yaw, and Roll.
- Quality detection: enabling --need-quality-detection returns Score, Sharpness, Brightness, and Completeness, helping decide whether an image is suitable for storage or further verification.
The typical flow is to provide image input, toggle attribute and quality options as needed, and receive JSON output. Compared with hand-writing SDK calls, this keeps parameters, defaults, and response fields in one skill entry point, making it easier to reuse in automated workflows.
Boundaries and Notes
This skill is suitable for image-level face detection, quality filtering, and attribute statistics. It is not a replacement for identity verification, 1:1 face matching, or long-term storage of sensitive biometric data. Input images must meet Tencent Cloud requirements: PNG, JPG, JPEG, and BMP are supported, Base64 size must not exceed 5MB, and resolution limits apply. Blurry images, heavy occlusion, or very small faces can be affected by --min-face-size and quality scores. Treat attributes and pose fields as probabilistic or heuristic signals, not deterministic facts.
Use Cases
- Check face-capture photos against enrollment thresholds by reading score, sharpness, brightness, and completeness.
- Generate face bounding boxes for avatar batches to crop, annotate, or verify whether a face is present.
- Count visible faces in group photos and return up to 120 face positions for downstream processing.
- Emit gender, age range, expression, and pose attributes to support rule-based photo review.
Best For
- Face-capture system engineers who need to validate enrollment photos and read quality and occlusion scores.
- Image review developers who need face boxes, face counts, and pose angles for rule-based pipelines.
- Automation test engineers who need batch checks that avatars contain faces and expected attribute fields.
- Photo platform engineers who need pre-checks on URL or Base64 images before further processing.
Related Skills
Guides Java Spring Boot backend development across REST APIs, data access, security, microservices, testing, deployment, and troubleshooting.
A systematic code error fixing skill that follows a five-step workflow: Reproduce, Isolate, Understand, Fix, Verify, to help developers resolve code errors.
Restores Lanhu designs to target framework code using schema JSON, design tokens, and image assets.
Automatically detects engineering or competition code, analyzes bugs, vulnerabilities, algorithm complexity, and edge cases, then produces risk-level findings with before/after fixes.