# Get started with object detection using a Jetson Orin Nano

## In this learning path

- [Introduction](https://learn.arm.com/learning-paths/embedded-and-microcontrollers/jetson_object_detection/)
- [Set up your Jetson Orin Nano](https://learn.arm.com/learning-paths/embedded-and-microcontrollers/jetson_object_detection/2setup/)
- [Launch the image classification Docker container](https://learn.arm.com/learning-paths/embedded-and-microcontrollers/jetson_object_detection/3docker/)
- [Detect objects in video and images](https://learn.arm.com/learning-paths/embedded-and-microcontrollers/jetson_object_detection/4object/)
- [Next Steps](https://learn.arm.com/learning-paths/embedded-and-microcontrollers/jetson_object_detection/_next-steps/)

## About this Learning Path

| Skill level: | Introductory |
|--------------|--------------|
| Reading time: | 1 hr |
| Last updated: | 13 Aug 2026 |

| Author: | Gabriel Peterson, Arm [GitHub](https://github.com/gabrieldpeterson) [LinkedIn](https://linkedin.com/in/gabrieldpeterson) [Twitter](https://twitter.com/@gabedpeterson) [Cortex Implant](https://corteximplant.com/@gabe) |
|-----------|----------|
| Arm IP: | [Cortex-A](https://support.arm.com/?tab=compute-ip&Product%20Type=Application%20Processors) |
| Tags: | ML, Linux, DetectNet, TensorRT, Docker |

### Who is this for?
This is an introductory topic for developers interested in integrating object detection into their applications.

### What will you learn?
Upon completion of this Learning Path, you will be able to:

- Set up a Jetson Orin Nano with a MIPI CSI-2 camera for object detection
- Detect objects from both live video and image files

### Prerequisites
Before starting, you will need the following:

- A [Jetson Orin Nano](https://developer.nvidia.com/embedded/learn/jetson-orin-nano-devkit-user-guide/index.html)
- A microSD card (64GB UHS-1 or larger is recommended)
- A MIPI CSI-2 camera, with a 22 pin connector on at least one end

### Summary
You’ll prepare a Jetson Orin Nano for object detection with a microSD image and MIPI CSI-2 camera. First, you’ll clone `jetson-inference`, launch its Docker container, and run TensorRT-accelerated DetectNet on live camera input and image files. Then, you’ll adjust the detection threshold and verify labeled objects in the resulting video and images.

### Frequently asked questions

<details>
<summary>Which image should I download for the microSD card?</summary>
On the NVIDIA developer website, expand **JETSON XAVIER NX DEVELOPER KIT & ORIN NANO DEVELOPER KIT**, then select **JETSON Orin Nano DEVELOPER KIT** to download the latest image.
</details>

<details>
<summary>Where should I run the Docker commands, and how do I get the container ID?</summary>
Run the Docker commands in a terminal on the host, not from inside the running container. To print the container ID, use `sudo docker ps -q`.
</details>

<details>
<summary>How do I start DetectNet on the live camera, and from which directory?</summary>
Change into the binaries directory with `cd build/aarch64/bin`. Start the live camera feed with `./detectnet csi://0`.
</details>

<details>
<summary>How can I adjust detection sensitivity, and what is the default?</summary>
Use the `--threshold` option to change sensitivity. For example, run `./detectnet csi://0 --threshold=0.25`. The default is `0.5`.
</details>

<details>
<summary>How do I run DetectNet on an image and save the annotated output?</summary>
From `build/aarch64/bin`, run `./detectnet --network=ssd-mobilenet-v2 images/peds_0.jpg` followed by your output path, such as `images/test/output.jpg`. The `--network` option is optional, and you can use `docker cp` to add your own images to the container.
</details>