How Machines Try to Understand Human Feelings?

Hi everyone, during my studies about machine learnings and digital humanities i found myself thinking more and more about a simple question:

Do you really think that machines or AI can understand us?

Maybe they can but only as much as we allow them to. As much as we define, label, and explain ourselves to them.

But what happens when machines see us instead of reading us?

I want to talk about my projects. I was working for my Visual Semiotics course and the topic was Face and Emotion Recognition by Machines and i did my project with both way :

First: Using already trained data which is OpenCv/ Deepface library.

Second: Using Gemini AI studio and i generate a Face and Emition Recognition App.

The Deep Learning Powerhouse: DeepFace

DeepFace is a library that wraps powerful models like VGG-Face, Google FaceNet, and OpenFace.

In the first stage of my project, I use DeepFace which is a powerful deep learning framework for Python. DeepFace operates by transforming facial features into a series of numerical vectors. From a Visual Semiotics standpoint, this is the ‘Denotative’ stage. The machine doesn’t know what ‘sadness’ feels like; it simply recognizes a specific alignment of facial landmarks and matches them with a label in its database.

So that means even machines can try to understand humans emotions with datas which is provide by humans.

DeepFace output interface via webcam

Crafting an AI Face Analyst with Gemini

While DeepFace focuses on the signifier (the physical facial muscles), Gemini attempts to understand the deeper meaning and context. By using prompt engineering in Gemini AI Studio, I didn’t just build a detector; I created an analyst

For the second part of my project, I started using Gemini AI Studio. This was a completely different experience. Instead of just looking at points on a face, I used Gemini’s ability to see and understand images like a human.

I gave Gemini a specific ‘prompt’ to act as a Professional Face Analyst. Traditional tools only look at the geometry of the face, but Gemini is much smarter. It looks at the whole picture to see emotions, gender, facial muscles moves and gives you summary of all.

Gemini AI Studio Face Analyst Interface

Data vs. Meaning

“So, can machines really understand our feelings?”

After working with both DeepFace and Gemini, I realized that there is a big difference between detecting an emotion and understanding it. DeepFace is excellent at reading the data on our faces the pixels and the math. But Gemini shows us the future: a machine that looks at the context and the ‘story’ behind the face.

As a computer engineer studying Digital Humanities, I believe the future of AI is not just about better code or better prompt. It is about teaching machines the ‘semiotics’ of being human. AI can now see our smiles, but the next challenge is for AI to understand what those smiles really mean in our complex human world.

Made by Gemini