Building a Video Intelligence Pipeline
A video intelligence pipeline for structured facial analytics.
Project statement
Turning video into structured human signals.
Context
Mental Edge began as a project within Innovated Studios and was spun out as a company in 2024.
Problem
The job is to turn image and video into structured facial analytics.
Constraints
Observable facial movement, facial analytics, and emotion analysis are different things — none of them is a read on what someone is thinking.
My role and personal contribution
I am a co-founder, and I personally trained the emotion-analysis model powering the current API.
System
Public capabilities include face detection, embeddings, recognition, video analysis, facial analytics, and an emotion-analysis API.
Decisions and tradeoffs
If the language slides into mind-reading, the work is wrong even when the model is useful.
Verified outcome
The current API is powered by the emotion-analysis model I trained.
Lesson
The system sees the face as it moves. That is not a claim about someone’s inner state.
Relevant domains
- Computer vision
- Video analysis
- Facial analytics
- Production AI