Computer vision / Early experiment
Site Safety Detection
Detect people, helmets, hi-vis vests and gloves in construction-site photographs.
- My contribution
- I fine-tuned the detector and assembled the recorded examples described in the research page.
- Tools & methods
- YOLO · Fine-tuning · Object detection
The problem
What this project investigates
Identifying equipment in an image involves small objects, occlusion and difficult lighting. This experiment explores those detection cases.
The build
What I built
The portfolio shows a fine-tuned YOLO detector drawing boxes around people and equipment in still photographs.
Technical choices
Decisions behind the build
Keep the task narrow
The demonstration focuses on four object classes.
Show missing equipment
Recorded examples include a person without detected protective equipment.
Current status
What the evidence shows
A recording is available, but no public precision, recall, mAP, model version or dataset documentation accompanies it. This is an experiment, not a validated safety-monitoring system.
Further work & evidence gaps
The source lists evaluation, failure-case analysis and video input as next steps.