1 of 11

AI Integration

acmcsuf.com/ai/integrating-ai-sp25

2 of 11

FullyHacks

FullyHacks is this weekend!

​

​

ig: @fullyhacks

3 of 11

Integrating AI

MediaPipe - Computer vision framework

Gemini AI - Google's multimodal AI

OpenAI Whisper - Speech-to-text (if time permits)

Jupyter Notebooks for development

Additional resources:

  • HuggingFace
  • TensorFlow
  • PyTorch

​

​

4 of 11

Mediapipe by Google

  • Open-source framework for machine learning pipelines
  • Cross-platform multimedia processing
  • Specializes in real-time computer vision:
    • Hand tracking
    • Face detection
    • Pose estimation
    • Object detection

​

​

​

5 of 11

Gemini by Google

  • Google's most capable multimodal AI model
  • Access through Google AI Studio (ai.google.dev)
  • Python API available with `google-generativeai` package
  • Free tier available (Gemini 1.0 Pro & 1.5 Flash)
  • Set up with API keys stored in environment variables
  • Great for text, image, and multimodal applications

6 of 11

Whisper by OpenAI

  • Open-source speech recognition model
  • Multilingual and multitask capabilities
  • Install with `pip install openai-whisper`
  • Base models available offline, no API key needed
  • Perfect for transcription and translation tasks
  • Can be integrated with other AI systems

7 of 11

Before we get started…

Check if Python is installed with `python –-version` or `python3 –-version` in the terminal (Depending on system, Python may use either `python` or `python3` as its name)

We’ll be working in a space known as a virtual environment.

Create a virtual environment with `python -m venv venv`, which makes a folder named `venv`, then activate it (The command to activate it differs depends on your system and shell; we’ll give it to you later).

You install Python packages with `pip install library_name`; if you’re in an activated virtual environment, then packages will only be installed in that virtual environment.

8 of 11

Clone this Repository

`git clone https://github.com/ryan-avancena/integrating-ai`

9 of 11

What we’re doing …

10 of 11

Gemini API Key

Whisper Documentation

AIzaSyA6Ijp_qN_rEl70Ob-J5G-1_CIyBsO7lrI

https://github.com/openai/whisper

11 of 11

Closing

Executive Board Applications are still open!

​

​