Skip to main content

5.2.1 Ultralytics Usage Guide

Last Version: 11/09/2025

Overview​

Ultralytics is a company focused on computer vision and deep learning, best known as a key developer and maintainer of the famous YOLO (You Only Look Once) series of object detection models. The name refers to both the team and their open-source Python library.

  • Core Product Ultralytics YOLO: A top open-source solution for object detection and image segmentation, supporting YOLOv3/v5/v8/v9/v11.

  • Main Features

    • Object Detection: Identifies and locates objects in images or videos.
    • Segmentation: Segments objects at the pixel level.
    • Pose Estimation: Detects human keypoints.
    • Classification: Performs image classification tasks.

  • Ecosystem Features

    • Pure Python implementation based on PyTorch.
    • Easy installation via pip install ultralytics.
    • Includes command-line tools and Python API out of the box.
    • Offers training, validation, inference, and export tools.
    • Supports deployment multiple formats like ONNX, TensorRT, CoreML, and OpenVINO (use ONNX for SpacemiT RISC-V).

More about Ultralytics, please visit Ultralytics Official Page

Application Areas​

  • Video surveillance and security
  • Autonomous driving and traffic analysis
  • Industrial inspection and robot vision
  • Medical image analysis
  • Smart retail

Framework Adaptation Notes​

Ultralytics uses PyTorch as its backend framework. Note: PyTorch currently does not leverage hardware acceleration on SpacemiT RISC-V platforms.

  • You can use Ultralytics on our platform for quick algorithm validation and prototyping.
  • For deployment with hardware acceleration, use the spacemit-ort framework.
  • Refer to the Demo Zoo and Model Quantization Guide for details on optimized deployment.

Environment Setup​

Install Dependencies​

sudo apt install python3-pip python3-venv libxrender1 libgl1 libglib2.0-0t64

Platform Requirements​

SpacemiT RISC-V Board: Must have the Bianbu ROS system image flashed.

Installing Ultralytics​

Set Up Sources​

pip config set global.index-url https://mirrors.aliyun.com/pypi/simple/
pip config set global.extra-index-url https://git.spacemit.com/api/v4/projects/33/packages/pypi/simple

Create and Activate a Virtual Environment​

python3 -m venv test1
source test1/bin/activate
pip install pip -U

Install Ultralytics​

pip install --prefer-binary ultralytics

Expected output after installation:

Test the Installation​

Sample Code Source: https://docs.ultralytics.com/zh/modes/predict/#plotting-results

Note: This example involves visualization. Please connect an HDMI screen, keyboard, and mouse to the board to run the commands locally.

  • Create a file named demo1.py with the following content:

    from PIL import Image

    from ultralytics import YOLO

    # Load a pretrained YOLO11n model
    model = YOLO("yolo11n.pt")

    # Run inference on 'bus.jpg'
    results = model(["https://ultralytics.com/images/bus.jpg", "https://ultralytics.com/images/zidane.jpg"])  # results list

    # Visualize the results
    for i, r in enumerate(results):
      # Plot results image
      im_bgr = r.plot()  # BGR-order numpy array
      im_rgb = Image.fromarray(im_bgr[..., ::-1])  # RGB-order PIL image

      # Show results to screen (in supported environments)
      r.show()

      # Save results to disk
      r.save(filename=f"results{i}.jpg")

  • Run:

    source test1/bin/activate
    python demo1.py

  • Expected Output: