Skip to content

Getting Started

Welcome to usls!

This guide will help you get up and running with the library in just a few minutes.

🚀 Start with YOLO Demo

Let's run the YOLO-Series demo to explore models with different tasks, precision and execution providers:

  • Tasks: detect, segment, pose, classify, obb
  • Versions: 5, 6, 7, 8, 9, 10, 11, 12, 13, 26
  • Scales: n, s, m, l, x
  • Precision (DType): fp32, fp16, q8, q4, q4f16, bnb4
  • Devices: cpu, cuda:0, tensorrt:0, coreml, openvino:CPU

First, clone the repository and navigate to the project root

git clone https://github.com/jamjamjon/usls.git
cd usls

Then, run the demo:

# Object detection with YOLO26n (FP16)
cargo run -r --example yolo -- --task detect --ver 26 --scale n --dtype fp16

# Requires "cuda-full" feature
cargo run -r -F cuda-full --example yolo -- --task segment --ver 11 --scale m --device cuda:0 --processor-device cuda:0

# Requires "tensorrt-full" feature
cargo run -r -F tensorrt-full --example yolo -- --device tensorrt:0 --processor-device cuda:0

# Requires "coreml" feature
cargo run -r -F coreml --example yolo -- --device coreml

For a full list of options, run:

cargo run -r --example yolo -- --help

📊 Performance Reference

Environment: NVIDIA RTX 3060Ti (CUDA 12.8) / Intel i5-12400F
Setup: YOLO26n, 640x640 resolution, COCO2017 val set (5,000 images)

EP Image
Processor		 DType Batch Preprocess Inference Postprocess Total
TensorRT CUDA FP16 1 ~233µs ~1.3ms ~14µs ~1.55ms
TensorRT-RTX CUDA FP32 1 ~233µs ~2.0ms ~10µs ~2.24ms
TensorRT-RTX CUDA FP16 1 ❓ ❓ ❓ ❓
CUDA CUDA FP32 1 ~233µs ~5.0ms ~17µs ~5.25ms
CUDA CUDA FP16 1 ~233µs ~3.6ms ~17µs ~3.85ms
CUDA CPU FP32 1 ~800µs ~6.5ms ~14µs ~7.31ms
CUDA CPU FP16 1 ~800µs ~5.0ms ~14µs ~5.81ms
CPU CPU FP32 1 ~970µs ~20.5ms ~14µs ~21.48ms
CPU CPU FP16 1 ~970µs ~25.0ms ~14µs ~25.98ms
TensorRT CUDA FP16 8 ~1.2ms ~6.0ms ~55µs ~7.26ms
TensorRT CPU FP16 8 ~18.0ms ~25.5ms ~55µs ~43.56ms

Multi-Batch Performance

When using a larger batch size (e.g., batch 8), CUDA Image processor significantly improves throughput on GPUs.

Next Steps

  • Installation

    Install usls in your own project

    Install →

  • Integration

    Learn how to integrate usls into your code

    Integrate →