Cambodia Traffic Signs Detection — Render (no Docker)

This folder deploys the existing Flask interface on Render's native Python runtime. It includes the MPWT logo, both model checkpoints, and a render.yaml blueprint. No Docker or UI rewrite is required.

Deploy from GitHub

1. Make a GitHub account if needed, then create a new public repository for this deployment. Since this bundle contains only the app and its two model checkpoints, upload the contents of this folder to the repository root; do not upload the whole research project or its datasets. Make sure render.yaml is at the repository root.
2. Create a free account at Render.
3. In the Render dashboard choose New → Blueprint and connect the GitHub repository. Render will read render.yaml; confirm the service plan is Free before creating it.
4. Wait for the first build and deploy. Share the public onrender.com address with colleagues.

Colleagues only need a web browser. Images use both models; video and webcam run detection only. Webcam access works in the hosted HTTPS page.

Free plan and privacy

• The service is public, has no password, and can be used by anyone with the link.
• Uploaded images and video frames are sent to Render for inference. The app does not save the original uploads. Do not upload private or sensitive material.
• Render's free web services sleep after 15 minutes without inbound traffic; waking the service can take about one minute. Free workspaces have 750 web-service instance hours per month. See Render's free-service limits.
• CPU inference may be slower. Build and try the service with representative images and videos first; if the free instance cannot load the models or respond reliably within its resource limits, a larger paid instance may be necessary.
• The model weights are included in this public repository, so they are public too. Anyone may download them.
• Stop or delete the service in Render when you no longer need it. Free-plan terms and included usage can change; check Render's current pricing and dashboard before deploying.

Deployment settings

render.yaml sets the service root to this folder, installs CPU-only PyTorch dependencies, checks /api/health, and starts Flask with Gunicorn bound to Render's $PORT. See Render's Flask deployment guide and Blueprint reference.