EmotionLens: Exploring AI-Powered Facial Expression Analysis with Open-Source AI

Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle Submission 💸

This is a submission for the "Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle" ( https://dev.to/challenges/hacktoberfest-week2-2026-10-12 )

I'm exploring how AI-powered facial expression analysis can become a practical tool for independent creators, presentation coaches, and trainers.

My project, AI Emotions Detection, is the starting point for EmotionLens, a product concept focused on helping users explore facial-expression predictions and review patterns in recorded presentations.

🔗 GitHub repository: https://github.com/pratikdevelop/AI-emotions-detection

The problem I'm interested in solving is straightforward: reviewing recordings manually can be time-consuming. An accessible AI-assisted tool could help users organize observations and identify moments they want to review more closely.

The longer-term vision is to develop a user-friendly application with expression timelines, session summaries, and privacy-conscious processing.

I see potential customers among independent creators, coaches, and trainers. Possible future monetization includes paid reporting features or a self-hosted version, although willingness to pay still needs validation.

Facial-expression predictions are imperfect and don't reliably establish someone's actual emotional state. The goal is to support review, not make definitive judgments about people.

GitHub repository: https://github.com/pratikdevelop/AI-emotions-detection

A deployed demo or video demonstration will be added when available.

https://github.com/pratikdevelop/AI-emotions-detection

I'm using this project as a foundation for exploring how AI applications can evolve into useful, privacy-conscious products.

The project explores AI-based facial emotion detection. My goal is to develop it into a more complete application with a clear user workflow.

These are development goals, not claims that every feature is already implemented. The repository documents the current starting point.

I believe open innovation gives developers more control over how AI applications are built, deployed, and improved.

For a project that may process personal recordings, local inference with a suitable open model could help keep video data on the user's own device rather than sending it to an external AI provider.

Open models also make experimentation more flexible. Developers can compare models, evaluate performance, and adapt their applications without depending entirely on a single provider's API pricing or availability.

Running models locally isn't automatically free or private: hardware requirements, application architecture, storage, and telemetry still matter. But open-source tools give developers more choices about these trade-offs.

That's the direction I want to explore with this project: accessible AI, greater user control, and a realistic path from experimentation to a useful product.

Potential category to target: Best Use of Gemma

I plan to investigate whether an appropriate Gemma open-weight model can add value through optional session summaries. I would only claim this category after implementing and demonstrating a meaningful integration.

Potential category to target: Best Use of Render

I would also consider Render for hosting the application once there is a working deployment that meaningfully uses the platform.

These are intended targets, not claims of completed partner integrations.

My next step is to turn this starting point into a more complete application, test it with potential users, and find out which features actually save them time.

I'd love feedback from developers working with open-weight models, computer vision, and privacy-conscious AI applications.

What would make an AI-assisted presentation-review tool useful enough for you to use regularly?