Оптимизируйте модели глубокого обучения с помощью TensorRT для более быстрых и эффективных вычислений на NVIDIA GPU. Добейтесь производительности в реальном времени с YOLO и приложениями ИИ.
TensorRT is a high-performance deep learning inference software development kit (SDK) developed by NVIDIA. It is designed to optimize neural network models for deployment, delivering low inference latency and high throughput for deep learning applications. By acting as an optimization compiler, TensorRT takes trained networks from popular frameworks like PyTorch and TensorFlow and restructures them to execute efficiently on NVIDIA GPUs. This capability is crucial for running complex AI models in production environments where speed and efficiency are paramount.
The core function of TensorRT is to convert a trained neural network into an optimized "engine" specifically tuned for the target hardware. It achieves this through several advanced techniques:
Благодаря своей способности обрабатывать огромные объемы данных с минимальной задержкой, TensorRT широко TensorRT в отраслях, основанных на компьютерном зрении и сложных задачах искусственного интеллекта, где время имеет решающее значение.
Интеграция TensorRT в ваш рабочий процесс - это простое решение с помощью современных инструментов искусственного интеллекта. Сайт ultralytics package
provides a seamless method to convert standard PyTorch models into TensorRT engines. This allows users to leverage the
state-of-the-art architecture of Ultralytics YOLO26 with the
hardware acceleration of NVIDIA GPUs. For teams looking to manage their datasets and training pipelines before export,
the Платформа Ultralytics offers a comprehensive environment to prepare
models for such high-performance deployment.
Следующий пример демонстрирует, как экспортировать модель YOLO26 в файл TensorRT (.engine) и
использовать его для выводы в режиме реального времени:
from ultralytics import YOLO
# Load the latest stable YOLO26 model (nano size)
model = YOLO("yolo26n.pt")
# Export the model to TensorRT format (creates 'yolo26n.engine')
# This step optimizes the computational graph for your specific GPU
model.export(format="engine")
# Load the optimized TensorRT engine for high-speed inference
trt_model = YOLO("yolo26n.engine")
# Run inference on an image source
results = trt_model("https://ultralytics.com/images/bus.jpg")
It is important to distinguish TensorRT from other terms often heard in the model deployment landscape:
For developers aiming to maximize the performance of their AI agents or vision systems, understanding the transition from a training framework to an optimized runtime like TensorRT is a key step in professional MLOps.