Uploaded November 2025 | Updated November 2026, 3 days ago
Join Roboflow Engineer Riaz Virani as he shares a fun and practical computer vision application, Hydrovisor, which validates the simple human action of drinking water for real-time, automated hydration tracking. We use this as an example to illustrate how computer vision applications can run in edge devices and assist with real-time action validation.
Chapters:
00:00 Introduction: AI-Powered Hydration Reminders
01:58 Demonstration: Hydrovisor - RF-DETR in a Web Browser
05:48 Background: Why Build the Hydrovisor App & Reviewing Open Source Tools
09:10 Parsing Actions for a Computer to Understand Them
17:37 Data Collection and Training a Custom Model to Recognize Cups
23:42 Validating Actions with AI in Industrial Settings - How Would You Start?
24:53 Deploying with Inference.js and Considerations for Edge Devices
28:34 Designing Computer Vision Solutions for Industrial Applications
36:30 Handling Latency and Re-Training the Model
40:49 Dataset Preparation Tips for Reducing False Positives
45:49 Conclusion
Read the blog!
blog.roboflow.com/drinking-detection
Join a future live session:
roboflow.com/webinar
Join Roboflow Engineer Riaz Virani as he shares a fun and practical computer vision application, Hydrovisor, which validates the simple human action of drinking water for real-time, automated hydration tracking. We use this as an example to illustrate how computer vision applications can run in edge devices and assist with real-time action validation.
Chapters:
00:00 Introduction: AI-Powered Hydration Reminders
01:58 Demonstration: Hydrovisor - RF-DETR in a Web Browser
05:48 Background: Why Build the Hydrovisor App & Reviewing Open Source Tools
09:10 Parsing Actions for a Computer to Understand Them
17:37 Data Collection and Training a Custom Model to Recognize Cups
23:42 Validating Actions with AI in Industrial Settings - How Would You Start?
24:53 Deploying with Inference.js and Considerations for Edge Devices
28:34 Designing Computer Vision Solutions for Industrial Applications
36:30 Handling Latency and Re-Training the Model
40:49 Dataset Preparation Tips for Reducing False Positives
45:49 Conclusion
Read the blog!
blog.roboflow.com/drinking-detection
Join a future live session:
roboflow.com/webinar










