Tools and Workflow
13 articles
Building a Replay Set of Recorded Footage
A small, curated library of recorded clips with known events lets you test every pipeline change against reality. What to include and how to keep it.
Cloud GPUs for Model Training: How to Choose
How to choose a cloud GPU for training vision models: compare cost models, GPU memory, storage and reliability, and avoid the habits that waste money.
Experiment Tracking: TensorBoard, Weights & Biases and MLflow
Why experiment tracking matters for computer vision, what to log on every training run, and how TensorBoard, Weights & Biases and MLflow compare.
Host Computer Vision Demos for Free on GitHub Pages
Host a computer vision demo free on GitHub Pages: static hosting, running an ONNX model in the browser, size limits, licensing and a portfolio demo.
Metrics and Logging for a Vision Service
Which numbers to collect from a live video analytics service, per camera and per stage, so you see problems before the client does, and how to log.
OpenCV Essentials Every Computer Vision Engineer Needs
The OpenCV functions you use daily: reading images and video, drawing boxes, colour spaces, resizing, transforms, and the pitfalls that cause silent bugs.
Managing Per-Camera Configuration at Scale
Zones, thresholds, schedules and streams for hundreds of cameras: how to layer defaults and overrides, validate them and keep them versioned.
A Pre-Deployment Testing Checklist for Vision Models
A pre-deployment testing checklist for computer vision models: real footage, edge cases, day and night, latency, alerts, monitoring and rollback.
Profiling a Python Vision Pipeline: Finding the Real Bottleneck
Before optimising the model, find where the time actually goes. Stage timers, sampling profilers and the usual hidden costs in Python video pipelines.
Setting Up a Clean Python Environment for Computer Vision
Set up a clean Python environment for computer vision: venv, pinned requirements, matching PyTorch to CUDA, and reproducible installs.
Regression Tests for a Vision Pipeline
Automated tests that catch broken pre-processing, export mismatches and alert-rule bugs before they reach a live camera, without flaky tolerances.
Training on Google Colab: Limits and Workarounds
Train a YOLO model on Google Colab: what the free tier is good for, session limits, saving checkpoints to Drive, resuming runs and when to move on.
Versioning Datasets and Models Together
When a deployed model misbehaves, you need to know exactly which data, labels, code and settings made it. A light system for tracing every model back.