Woman Close Up

IoT & Hardware

IoT & Hardware

Bridging Computer Vision and Microcontrollers

Bridging Computer Vision and Microcontrollers

Bridging Computer Vision and Microcontrollers

Architecting a two-tier edge pipeline combining Raspberry Pi YOLOv8 neural models with ESP32 sensor relays.

Architecting a two-tier edge pipeline combining Raspberry Pi YOLOv8 neural models with ESP32 sensor relays.

Date:

Author:

Jarib Sioco

001

Segregating Brain and Brawn

Microcontrollers cannot handle neural inference, and single-board computers are poorly suited for high-reliability sensor polling and high-voltage AC contactor triggering. A robust edge deployment requires strict separation of concerns.

Woman Orange
Shoe

002

The Edge Vision Brain

A Raspberry Pi 4 Model B runs quantized YOLOv8 models locally over pond sluice gates. By tallying fingerlings non-invasively through camera feeds, we eliminated the severe physical netting stress responsible for high stock mortality.

Man Red BG

003

The Dedicated Actuator Brawn

An ESP32 microcontroller continuously polls submersible analog pH and dissolved oxygen probes. When DO drops beneath safety thresholds, hardware-isolated 5V relays engage paddlewheel aerators automatically.

Team

004

Cyber-Physical Cohesion

By piping telemetry data to a real-time Supabase dashboard over WebSockets, aquaculture operators receive immediate farm-wide visibility and emergency manual overrides from anywhere.

(01)

(Article)

© 2025

(01)

(Article)

(09)

(Frequently Asked Questions)

What can you build?

Do you design the architecture first?

Can you work with hardware and software?

What makes your AI workflow different?

Are you available for a serious build?

(09)

(Frequently Asked Questions)

What can you build?

Do you design the architecture first?

Can you work with hardware and software?

What makes your AI workflow different?

Are you available for a serious build?

(09)

(Frequently Asked Questions)

What can you build?

Do you design the architecture first?

Can you work with hardware and software?

What makes your AI workflow different?

Are you available for a serious build?