Computer Vision from the Edge to the Cloud done easy

Run powerful cloud AI without streaming endless video. This architecture uses smart cameras to trigger analysis on a single snapshot, slashing costs.

#1about 6 minutes

Defining computer vision and its real-world applications

Computer vision enables computers to understand digital images and videos, with applications in retail, public safety, traffic monitoring, and smart cities.

#2about 2 minutes

Understanding the key components of a vision system

A typical computer vision architecture includes cameras for recording, storage for footage, a machine learning pipeline for processing, and a dashboard for results.

#3about 3 minutes

Exploring the features of Cisco Meraki IP cameras

Cisco Meraki cameras are cloud-managed devices with on-board storage and processing for detecting people, vehicles, and audio events like fire alarms.

#4about 2 minutes

Integrating cameras using APIs, MQTT, and RTSP streams

Meraki cameras offer multiple integration points including a REST API, webhooks for cloud events, local MQTT for real-time triggers, and RTSP for video streaming.

#5about 6 minutes

Demoing real-time event detection and analysis

A live demonstration shows how a camera's local MQTT broker can trigger events for person detection in a zone and audio alarm recognition.

#6about 5 minutes

Designing an efficient event-driven vision architecture

Use on-camera analytics and MQTT triggers to send a single snapshot to a cloud vision API for analysis, reducing bandwidth and processing costs.

#7about 2 minutes

Comparing pre-trained models from AWS, Azure, and GCP

A comparison of the pre-trained computer vision models and pricing tiers available on AWS Rekognition, Azure Computer Vision, and Google Cloud Vision API.

#8about 1 minute

Deciding between pre-trained and custom vision models

While pre-trained models are easy to use, building a custom model with your own dataset is necessary for highly specific detection tasks.

#9about 3 minutes

Showcasing computer vision project examples

Practical examples demonstrate architectures for detecting face masks, capturing license plates, and using door sensors to trigger snapshots for analysis.

#10about 18 minutes

Answering audience questions on practical implementation

The Q&A session covers topics like using cameras for home security, filtering out pets from alerts, and the challenges of creating custom models for specific tasks.

From learning to earning

Jobs that call for the skills explored in this talk.