CV Brief · Monday, 12 October 2026
CV Brief
Tutorials & Guides
Defect Detection with LLM-as-Judge: Production Quality Control
Explores using LLMs as judgment layer for defect detection pipelines, automating quality control decisions. The approach leverages pre-existing model capabilities through output channel optimization rather than retraining. Directly applicable to manufacturing inspection systems where LLM confidence scoring can replace hand-crafted thresholds.
Read more →Document Edge Detection on Mobile: Benchmark SmartCropper vs YOLO
Benchmarks three edge detection approaches (SmartCropper, MobileNetV3, YOLO) for on-device document scanning with quantified error rates. Provides practical model comparison for production mobile scanning pipelines. Critical for anyone optimizing document capture systems with latency/accuracy tradeoffs.
Read more →Getting Started in CV/ML
Fine-Tuning Upscaler Models: Semiconductor Image Enhancement
Case study on adapting existing upscaling models for semiconductor chip imagery through targeted fine-tuning. Demonstrates practical transfer learning workflow from hackathon to deployment. Key for practitioners needing domain-specific image enhancement without training from scratch.
Read more →Auto-labeling confidence threshold: don't use 0.5. For quality training data, start at 0.7 and manually review the 0.5–0.7 band. The borderline cases are where your model learns.