I’ve been working on a lightweight computer vision app that recognizes PPE in images and videos. Upload an image or video, and it returns annotated outputs plus the detected class labels (e.g., helmet, vest, gloves). Perfect for quick demos, audits, or integrating into ops dashboards.
What it does:
Detects 6 PPE classes in images and uploaded videos
Shows annotated frames and a clean list of detected classes
Streams annotated video via MJPEG for near real-time review
Tech stack:
Python 3.10, Flask (single-file server app.py)
Ultralytics YOLOv8, PyTorch (CPU-only option available)
OpenCV, Pillow, NumPy
Simple HTML/JS UI embedded in the Flask route
REST endpoints: /predict, /predict_image, /upload_video, /stream/<id>
What it’s useful for:
Safety compliance checks (warehouses, construction, manufacturing)
Faster visual audits and incident reviews
Operator dashboards that surface missing PPE
Prototyping edge-ready CV pipelines (CPU-friendly path included)
Related: Built a real-time Pipe Detection system with YOLOv8.
Related: I Built an AI Instagram Carousel Generator in One Weekend.
Discover more from Susiloharjo
Subscribe to get the latest posts sent to your email.