Human Visual Emotion Recognition Analysis
Paste the following prompt into your AI chat to install this skill:
Please install @user_bb47e3e3/human-emotion-recognition-analysis according to https://skillhub.cn/install/skillhub.md.
About this skill
Problem
In interactive, customer-support, or wellness-monitoring contexts, a user may provide only an image or short video, while the missing signal is an interpretable estimate of affect. Manual observation cannot reliably cover dimensions such as happiness, sadness, depression, calmness, anger, surprise, and fear, and it is hard to judge intensity shifts or abnormal segments.
How it works
The skill accepts frontal-face images or videos through --input or --url, then uses scripts/human_emotion_recognition_analysis.py to call the analysis service. Its core capabilities include:
- Multi-class emotion recognition: identifies common affect dimensions from frontal facial cues;
- Intensity quantification: outputs emotion scores and can flag anomalies with --threshold;
- Structured reporting: produces results, risk notes, suggestions, and report links;
- History lookup: reads historical reports from the cloud via --list, avoiding local-memory lookups.
When running the workflow, keep the command context in the skill root, set the media type with --media-type video/image, and adjust output detail with --detail json.
Boundaries
The output is a state reference for interaction or monitoring, not a professional psychological consultation or diagnosis. Inputs should be clear and suitable for the use case; videos support mp4/avi/mov, images support jpg/png/jpeg, and files must stay under 10MB. Persistent abnormal emotion should be escalated to professional help.
Use Cases
- After customer-service interactions, use provided chat-recording images to generate happiness, anger, and calm emotion reports with intensity scores.
- In mental-health services, mark threshold-based emotion anomalies in client-provided selfie video clips and compile risk notes for professional review.
- During human-computer interaction testing, score multi-class facial emotions in frontal-face prototype videos to build comparable trend materials.
- During report review, query cloud historical emotion reports by date range and locate abnormal intensity samples from previous runs.
Best For
- Customer-service QA leads reviewing call recordings who need structured reports of emotion dimensions and abnormal intensity.
- Mental-health monitoring product engineers who need intensity scores and anomaly flags from facial video clips to support alerting rules.
- Human-computer interaction researchers comparing multi-class frontal-face emotion distributions and trends across prototype demos.
- Analytics operations staff reviewing historical reports who need cloud report lists and abnormal sample lookup.
Related Skills
An AI workflow that structures personal knowledge bases, IMA, Yuque, and Feishu content into approved personas for paid WeChat mini-agent deployment.
Local workflow memory with matching and SOP updates.
An OpenClaw live streaming executor that initializes TRTC streaming, starts a real-time dashboard, generates viewer URLs, and continuously reports live events.
Breaks down physical supply chains for super-trends to identify second- and third-layer bottlenecks, runs valuation and reverse checks, and maintains trackable reports.