BoumedineBillal

BoumedineBillal / yolo26n_esp

Public

World's First NMS-Free YOLOv26n on ESP32-P4. Features end-to-end Int8 QAT and custom C++ optimizations achieving 30% faster inference than the official ESP-DL YOLOv11n (1.7s vs 2.4s).

boumedinebillal.github.iomy_profileproject-viewer.html?id=yolo26n-esp32p4 computer-vision edge-ai embedded-ai esp-dl esp-idf
13
0
100% credibility
Found Feb 06, 2026 at 10 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Jupyter Notebook
AI Summary

This repository offers a complete workflow to optimize, train, and deploy a high-speed object detection model on the ESP32-P4 microcontroller for edge devices.

How It Works

1
🕵️ Discover the project

You hear about a way to make a tiny gadget like ESP32-P4 spot objects in photos super fast, like buses and people.

2
📥 Grab the files

Download the project folder to your computer to get started with everything you need.

3
🛠️ Set up your workspace

Add a few helper programs to your computer so it can prepare the smart detector.

4
🎓 Train the detector

Open the easy guidebook on your computer and run it to teach the AI how to recognize objects accurately and quickly.

5
🔧 Prepare your gadget

Connect your small ESP32-P4 device and load the trained brain onto it with simple steps.

6
🚀 Run the magic

Feed it photos and watch as it instantly spots and outlines objects like people or buses in under 2 seconds.

🎉 Enjoy fast detections

Your tiny gadget now runs powerful object spotting on its own, perfect for smart cameras or robots.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 10 to 13 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is yolo26n_esp?

This repo delivers the world's first NMS-free YOLOv26n model running on ESP32-P4, slashing inference to 1.7s via end-to-end Int8 quantization-aware training and custom C++ optimizations. Developers get a Jupyter Notebook pipeline to train on COCO, export quantized models, and flash ready-to-run ESP-IDF firmware for edge object detection. It matches YOLOv11n accuracy at 36.5% mAP but beats official ESP-DL benchmarks by 30%.

Why is it gaining traction?

NMS-free design skips heavy post-processing, pairing with hardware-tuned kernels for real 1.7s latency on low-power ESP32-P4—faster than stock YOLOv11n's 2.4s. Dual-res support (512x512 or 640x640) and seamless Python-to-firmware workflow hook devs chasing efficient vision without accuracy loss. Like world map github tools, it simplifies deploying production-grade detection.

Who should use this?

Embedded engineers building IoT cameras or battery-powered vision nodes on ESP32-P4. Ideal for surveillance prototypes, robot obstacle avoidance, or smart sensors where sub-2s inference matters over cloud dependency. Skip if you're not on Espressif hardware.

Verdict

Grab it if ESP32-P4 edge AI is your jam—docs are solid, benchmarks verified, and setup is straightforward with ESP-IDF v5.4+. At 12 stars and 1.0% credibility, it's early but tested; star it to push segmentation ports forward. (187 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.