https://github.com/roboflow/server-benchmark
A script you can use to benchmark the Roboflow Deploy targets with your custom trained model on your hardware.
https://github.com/roboflow/server-benchmark
benchmark computer-vision inference object-detection roboflow
Last synced: 10 months ago
JSON representation
A script you can use to benchmark the Roboflow Deploy targets with your custom trained model on your hardware.
- Host: GitHub
- URL: https://github.com/roboflow/server-benchmark
- Owner: roboflow
- License: apache-2.0
- Created: 2021-10-29T18:37:34.000Z (about 4 years ago)
- Default Branch: main
- Last Pushed: 2022-03-02T19:30:58.000Z (almost 4 years ago)
- Last Synced: 2025-02-25T09:59:40.615Z (11 months ago)
- Topics: benchmark, computer-vision, inference, object-detection, roboflow
- Language: JavaScript
- Homepage: https://docs.roboflow.com/inference
- Size: 4.17 MB
- Stars: 1
- Watchers: 3
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Roboflow Server Benchmark Tool
This simple tool will fire a series of inference requests at the
[Roboflow remote inference API](https://docs.roboflow.com/inference/hosted-api)
or one of
[our Docker containers](https://hub.docker.com/r/roboflow/inference-server/tags)
running on your hardware and measures the throughput.
## Requirements
A recent version of Node.js (eg 12 or higher).
## Installation
`cd` into this directory and run `npm install`.
Optional: place your test images in the `images` folder; the repo contains some
images from the [EgoHands dataset](https://universe.roboflow.com/brad-dwyer/egohands-public)
## Usage
Edit `benchmark.js` to add your API Key and model endpoint then configure which
deployment target you're testing (and its local IP address if applicable).
Then run `node benchmark.js`
## Alternative Authentication
You can also set a ROBOFLOW_KEY environment variable or put your API key into a
`.roboflow_key` file in this directory.