Package Delivery Detection with ESP-Based Computer Vision

Reporting from Shanghai, China
Sep 8, 2022

Identify when packages or boxes arrive on your doorstep with this computer-vision project.

The expansion of e-commerce has resulted in an ever increasing number of people making online purchases and having their goods delivered on their doorstep. This convenience, however, also increases the potential of stealing, as thieves may sometimes be much quicker in “collecting" such parcels than their rightful owners.

Package Delivery Monitoring with TinyML

To prevent package theft, we shall use Edge Impulse for creating a Machine Learning model that recognizes parcels. The model will then be deployed on Espressif’s powerful ESP-EYE development board that has been made for image recognition and audio processing, in particular. It features an ESP32 chip, a 2-Megapixel camera and a microphone. ESP-EYE offers plenty of storage, with an 8 Mbyte PSRAM and a 4 Mbyte flash. It also supports image transmission via Wi-Fi and debugging through a Micro-USB port.

We will use has the 2MP camera of ESP-EYE, in order to collect the live video feed of our shipments accurately and on time.

To develop our Machine Learning model, we will use FOMO (Faster Objects, More Objects). This is an algorithm developed by Edge Impulse to enable real-time object detection, tracking and counting on microcontrollers. FOMO is 30x faster than MobileNet SSD and runs on <200K of RAM.

Quick Start

The software to be used, along with the ESP-EYE development board, is the Edge Impulse Studio. You can find the public project here: Parcel Detection - FOMO. To add this project into your account, click “Clone this project” at the top of the page. Next, go to the “Deploying ESP-EYE” section below, in order to learn how to deploy the model on the ESP-EYE board.

Alternatively, to create a similar project, you can follow the specific steps described in this Edge Impulse project.

Share this article
  • LinkedIn
  • 微信


Reuse this content


Technical Writer and Editor

About this author ›