You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
<divalign="center"><imgalt="original image"src="resources/isaac_ros_object_detection_example.png"width="300px"/> <imgalt="bounding box predictions using DetectNet"src="resources/isaac_ros_object_detection_example_bbox.png"width="300px"/></div>
4
4
5
5
## Overview
6
6
7
-
This repository provides a GPU-accelerated package for object detection based on [DetectNet](https://developer.nvidia.com/blog/detectnet-deep-neural-network-object-detection-digits/). Using a trained deep-learning model and a monocular camera, the `isaac_ros_detectnet` package can detect objects of interest in an image and provide bounding boxes. DetectNet is similar to other popular object detection models such as YOLOV3, FasterRCNN, SSD, and others while being efficient with multiple object classes in large images.
7
+
Isaac ROS Object Detection contains an ROS 2 package to perform object detection. `isaac_ros_detectnet` provides a method for spatial classification using bounding boxes with an input image. Classification is performed by a GPU-accelerated [DetectNet](https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/pretrained_detectnet_v2) model. The output prediction can be used by perception functions to understand the presence and spatial location of an object in an image.
8
+
9
+
<divalign="center"><imgalt="graph of nodes using DetectNet"src="resources/isaac_ros_object_detection_nodegraph.png"width="500px"/></div>
10
+
11
+
`isaac_ros_detectnet` is used in a graph of nodes to provide a bounding box detection array with object classes from an input image. A [DetectNet](https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/pretrained_detectnet_v2) model is required to produce the detection array. Input images may need to be cropped and resized to maintain the aspect ratio and match the input resolution of DetectNet; image resolution may be reduced to improve DNN inference performance, which typically scales directly with the number of pixels in the image. `isaac_ros_dnn_image_encoder` provides a DNN encoder to process the input image into Tensors for the DetectNet model. Prediction results are clustered in the DNN decoder to group multiple detections on the same object. Output is provided as a detection array with object classes.
12
+
13
+
DNNs have a minimum number of pixels that need to be visible on the object to provide a classification prediction. If a person cannot see the object in the image, it’s unlikely the DNN will. Reducing input resolution to reduce compute may reduce what is detected in the image. For example, a 1920x1080 image containing a distant person occupying 1k pixels (64x16) would have 0.25K pixels (32x8) when downscaled by 1/2 in both X and Y. The DNN may detect the person with the original input image, which provides 1K pixels for the person, and fail to detect the same person in the downscaled resolution, which only provides 0.25K pixels for the person.
14
+
15
+
> **Note**: DetectNet is similar to other popular object detection models such as YOLOV3, FasterRCNN, and SSD, while being efficient at detecting multiple object classes in large images.
16
+
17
+
<divalign="center"><imgalt="comparison of bounding box detection to segmentation"src="resources/isaac_ros_object_detection_example_bboxseg.png"width="300px"/></div>
18
+
19
+
Object detection classifies a rectangle of pixels as containing an object, whereas image segmentation provides more information and uses more compute to produce a classification per pixel. Object detection is used to know if, and where in a 2D image, the object exists. If a 3D spacial understanding or size of an object in pixels is required, use image segmentation.
20
+
21
+
### DNN Models
22
+
23
+
To perform DNN inferencing a DNN model is required. NGC provides [DetectNet pre-trained models](https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/pretrained_detectnet_v2) for use in your robotics application. Using [TAO](https://developer.nvidia.com/tao-toolkit) these pre-trained models can be fine-tuned for your application.
8
24
9
25
### Isaac ROS NITROS Acceleration
10
26
@@ -18,20 +34,20 @@ The performance results of benchmarking the prepared pipelines in this package o
> **Note:** These numbers are reported with defaults parameter values found in [params.yaml](./isaac_ros_detectnet/config/params.yaml).
37
+
> **Note**: These numbers are reported with defaults parameter values found in [params.yaml](./isaac_ros_detectnet/config/params.yaml).
22
38
23
39
These data have been collected per the methodology described [here](https://github.com/NVIDIA-ISAAC-ROS/.github/blob/main/profile/performance-summary.md#methodology).
24
40
25
-
### ROS2 Graph Configuration
41
+
### ROS 2 Graph Configuration
26
42
27
-
To run the DetectNet object detection inference, the following ROS2 nodes should be set up and running:
43
+
To run the DetectNet object detection inference, the following ROS 2 nodes should be set up and running:
28
44
29
45
![DetectNet output image showing 2 tennis balls correctly identified](resources/ros2_detectnet_node_setup.svg"Tennis balls detected in image using DetectNet")
30
46
31
47
1.**Isaac ROS DNN Image encoder**: This will take an image message and convert it to a tensor ([`TensorList`](https://github.com/NVIDIA-ISAAC-ROS/isaac_ros_common/blob/main/isaac_ros_tensor_list_interfaces/msg/TensorList.msg) that can be
32
48
processed by the network.
33
49
2.**Isaac ROS DNN Inference - Triton**: This will execute the DetectNet network and take as input the tensor from the DNN Image Encoder.
34
-
> **Note:** The [Isaac ROS TensorRT](https://github.com/NVIDIA-ISAAC-ROS/isaac_ros_dnn_inference/tree/main/isaac_ros_tensor_rt) package is not able to perform inference with DetectNet models at this time.
50
+
> **Note**: The [Isaac ROS TensorRT](https://github.com/NVIDIA-ISAAC-ROS/isaac_ros_dnn_inference/tree/main/isaac_ros_tensor_rt) package is not able to perform inference with DetectNet models at this time.
35
51
36
52
The output will be a TensorList message containing the encoded detections. Use the parameters `model_name` and `model_repository_paths` to point to the model folder and set the model name. The `.plan` file should be located at `$model_repository_path/$model_name/1/model.plan`
37
53
3.**Isaac ROS Detectnet Decoder**: This node will take the TensorList with encoded detections as input, and output `Detection2DArray` messages for each frame. See the following section for the parameters.
@@ -40,9 +56,10 @@ To run the DetectNet object detection inference, the following ROS2 nodes should
@@ -64,24 +81,24 @@ To run the DetectNet object detection inference, the following ROS2 nodes should
64
81
65
82
## Latest Update
66
83
67
-
Update 2022-10-19: Updated OSS licensing
84
+
Update 2023-04-05: Source available GXF extensions
68
85
69
86
## Supported Platforms
70
87
71
-
This package is designed and tested to be compatible with ROS2 Humble running on [Jetson](https://developer.nvidia.com/embedded-computing) or an x86_64 system with an NVIDIA GPU.
88
+
This package is designed and tested to be compatible with ROS 2 Humble running on [Jetson](https://developer.nvidia.com/embedded-computing) or an x86_64 system with an NVIDIA GPU.
72
89
73
-
> **Note**: Versions of ROS2 earlier than Humble are **not** supported. This package depends on specific ROS2 implementation features that were only introduced beginning with the Humble release.
90
+
> **Note**: Versions of ROS 2 earlier than Humble are **not** supported. This package depends on specific ROS 2 implementation features that were only introduced beginning with the Humble release.
| Jetson |[Jetson Orin](https://www.nvidia.com/en-us/autonomous-machines/embedded-systems/jetson-orin/) <br> [Jetson Xavier](https://www.nvidia.com/en-us/autonomous-machines/embedded-systems/jetson-agx-xavier/)|[JetPack 5.0.2](https://developer.nvidia.com/embedded/jetpack)| For best performance, ensure that [power settings](https://docs.nvidia.com/jetson/archives/r34.1/DeveloperGuide/text/SD/PlatformPowerAndPerformance.html) are configured appropriately. |
| Jetson |[Jetson Orin](https://www.nvidia.com/en-us/autonomous-machines/embedded-systems/jetson-orin/) <br> [Jetson Xavier](https://www.nvidia.com/en-us/autonomous-machines/embedded-systems/jetson-agx-xavier/)|[JetPack 5.1.1](https://developer.nvidia.com/embedded/jetpack)| For best performance, ensure that [power settings](https://docs.nvidia.com/jetson/archives/r34.1/DeveloperGuide/text/SD/PlatformPowerAndPerformance.html) are configured appropriately. |
To simplify development, we strongly recommend leveraging the Isaac ROS Dev Docker images by following [these steps](https://github.com/NVIDIA-ISAAC-ROS/isaac_ros_common/blob/main/docs/dev-env-setup.md). This will streamline your development environment setup with the correct versions of dependencies on both Jetson and x86_64 platforms.
83
100
84
-
> **Note:** All Isaac ROS Quickstarts, tutorials, and examples have been designed with the Isaac ROS Docker images as a prerequisite.
101
+
> **Note**: All Isaac ROS Quickstarts, tutorials, and examples have been designed with the Isaac ROS Docker images as a prerequisite.
85
102
86
103
## Quickstart
87
104
@@ -100,6 +117,10 @@ To simplify development, we strongly recommend leveraging the Isaac ROS Dev Dock
Copy file name to clipboardExpand all lines: docs/tutorial-isaac-sim.md
+10-11Lines changed: 10 additions & 11 deletions
Original file line number
Diff line number
Diff line change
@@ -6,6 +6,8 @@
6
6
7
7
This tutorial walks you through a pipeline for object(people) detection using [DetectNet](https://github.com/NVIDIA-ISAAC-ROS/isaac_ros_object_detection) consuming images from Isaac Sim.
8
8
9
+
Last validated with [Isaac Sim 2022.2.1](https://docs.omniverse.nvidia.com/app_isaacsim/app_isaacsim/release_notes.html#id1)
10
+
9
11
## Tutorial Walkthrough
10
12
11
13
1. Complete the [Quickstart section](../README.md#quickstart) in the main README.
@@ -39,16 +41,13 @@ This tutorial walks you through a pipeline for object(people) detection using [D
39
41
```
40
42
41
43
6. Install and launch Isaac Sim following the steps in the [Isaac ROS Isaac Sim Setup Guide](https://github.com/NVIDIA-ISAAC-ROS/isaac_ros_common/blob/main/docs/isaac-sim-sil-setup.md)
42
-
7. Open up the Isaac ROS Common USD scene (using the "content" window) located at:
0 commit comments