Let's add your hardware to Deeplite YOLOBench!

Interested in showcasing your hardware on the Deeplite YOLOBench app? 

Generating your own YOLOBench model files (in ONNX or TFLite) on your hardware platform is as simple as executing this script from Deeplite's GitHub Torch Zoo.  Once the model files are generated in the necessary format, they can be used for latency measurements on your hardware using your corresponding inference engine.

We recommend running several trials, each consisting of several inference passes (including several warmup passes), and then using the average latency value over trials to ensure a good representative sampling for your benchmarks.

Once you have gathered the values for all models, create a CSV file in this format and you can either:

1. Complete this form and upload your CSV file, or
2. Open a pull request in the Deeplite Torch Zoo repository and add the CSV file, or
3. Email the data to us directly at yolobench@deeplite.ai 

We also encourage you to share a product data sheet, marketing collateral or any other document you'd like us to make available to YOLOBench users! We will contact you when we've completed the process of adding your benchmark data to YOLOBench and we'll announce to the Community!  

Know someone who will be interested?

Share this content on: