Ultralytics
guides/deepstream-nvidia-jetson/ #14233
giscus[bot]
bot
Replies: 26 comments 105 replies
-
|
I have just done everything as I've seen, apparently i can't seem to run inference: Quitting |
Beta Was this translation helpful? Give feedback.
-
|
@AgbajeAyomipo hi there! Thank you for reaching out. It looks like you're encountering an issue with parsing the Here are a few steps to help troubleshoot and resolve the issue:
Here’s a quick example to ensure your setup is correct: # Clone the repository
git clone https://github.com/marcoslucianops/DeepStream-Yolo
cd DeepStream-Yolo
# Convert YOLOv8 model to ONNX
python3 utils/export_yoloV8.py -w yolov8s.pt --opset 12 --simplify
# Compile the library
make -C nvdsinfer_custom_impl_Yolo clean && make -C nvdsinfer_custom_impl_Yolo
# Edit config files as needed
# Run inference
deepstream-app -c deepstream_app_config.txtIf you follow these steps and still encounter issues, please share more details or any specific error messages you receive. This will help us assist you more effectively. Best of luck, and feel free to reach out if you have any more questions! 😊 |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for the reply, i fixed it already I was not in the directory when i ran the command apparently. |
Beta Was this translation helpful? Give feedback.
-
|
Hi @AgbajeAyomipo, Thank you for the update! I'm glad to hear that you were able to resolve the issue by running the command from the correct directory. It's a common oversight, and it happens to the best of us. 😊 If you have any further questions or run into any other issues, feel free to reach out. We're here to help! |
Beta Was this translation helpful? Give feedback.
-
|
is the ultralytics package necessary for following this guide? would just the .pt file be sufficient for exporting with deepstream-yolo |
Beta Was this translation helpful? Give feedback.
-
|
what does step 6 do? |
Beta Was this translation helpful? Give feedback.
-
|
since the onnx file has already been generated in a previous step |
Beta Was this translation helpful? Give feedback.
-
|
Step 6 involves compiling the custom YOLO inference library for DeepStream. The |
Beta Was this translation helpful? Give feedback.
-
|
@glenn-jocher, do we have to compile again each time there is a new onnx file? |
Beta Was this translation helpful? Give feedback.
-
|
Yes, you need to compile again each time there is a new ONNX file to ensure the custom YOLO layers are correctly integrated for inference. |
Beta Was this translation helpful? Give feedback.
-
|
Hello everyone, I'm trying to run the deepstream-app with an FP16 YOLOv8 model. I first followed this guide https://github.com/marcoslucianops/DeepStream-Yolo/blob/master/docs/YOLOv8.md but couldn't find an "fp16 option". Then, I tried to export a custom model with model.export(half=True,...), but the deepstream-app runs without displaying rectangles (I did set network-mode=2 in the config file). Has anyone successfully used an FP16 YOLO model with DeepStream? |
Beta Was this translation helpful? Give feedback.
-
|
@pderrenger by exporting the FP16 model you mean with model.export() or with this guide https://github.com/marcoslucianops/DeepStream-Yolo/blob/master/docs/YOLOv8.md ? I don't know how to do that with the DeepStream-Yolo guide so I suppose by running this : I also tried to directly export an |
Beta Was this translation helpful? Give feedback.
-
|
@metapole, you're right, I mixed up |
Beta Was this translation helpful? Give feedback.
-
|
you shouldn't be using that export if you want it to work with deepstream. the only difference is changing the config for fp16. for batching you would have to go beyond editing the inference component of the pipeline and understand how the end-to-end pipeline is set up. i personally use https://github.com/prominenceai/deepstream-services-library which is beginner friendly |
Beta Was this translation helpful? Give feedback.
-
|
Perfect it works now ! Yeah I thought it was necessary to import a "FP16 pre-processed" |
Beta Was this translation helpful? Give feedback.
-
|
Great to hear it's working now! For optimizing FPS, ensure your TensorRT engine is correctly set to FP16 by using |
Beta Was this translation helpful? Give feedback.
-
|
Hello everyone, I have just done everything as I've seen, and everything works fine. My problem is how to run RTSP streaming. I created one using VLC on my local Windows and modified these 2 arguments. type=4 but it's giving me a black screen and showing this error. |
Beta Was this translation helpful? Give feedback.
-
|
@seifeldin33 it seems like the RTSP URL might not be valid or accessible. Please ensure that the RTSP stream is correctly set up and accessible at |
Beta Was this translation helpful? Give feedback.
-
|
I have a question about integrating Kafka with Deepstream. My application is designed to send safety violations to the backend in real-time. I've heard about Kafka, and I'm wondering if it's the right approach to use Kafka for this purpose. I would also like to know how to integrate it. I want to ensure that the violations are sent securely in real-time. |
Beta Was this translation helpful? Give feedback.
-
|
@seifeldin33 using Kafka for real-time safety violation alerts is a solid choice due to its scalability and reliability. To integrate Kafka with DeepStream, you can use a Kafka producer to send messages from your application to a Kafka topic. Ensure your DeepStream pipeline processes the data and sends it to Kafka securely. For detailed integration steps, consider exploring Kafka's official documentation or community resources. |
Beta Was this translation helpful? Give feedback.
-
|
and what about Rest API? |
Beta Was this translation helpful? Give feedback.
-
|
For integrating a REST API with DeepStream, you can use the DeepStream SDK's message broker capabilities to send data to a REST endpoint. This involves configuring the message broker to use HTTP as the protocol. For detailed steps, refer to the DeepStream documentation. |
Beta Was this translation helpful? Give feedback.
-
|
Hi, I also have a separate venv with yolov8 to train on a custom dataset. My question is: can I just update this to yolo11 and next move my trained .pt file to the deepstream installation. Will this work? |
Beta Was this translation helpful? Give feedback.
-
|
The |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for the swift replies Glenn. |
Beta Was this translation helpful? Give feedback.
-
|
You're welcome! For your camera resolution, use |
Beta Was this translation helpful? Give feedback.
-
|
What is the opset for deepstream 6.2 Glenn? |
Beta Was this translation helpful? Give feedback.
-
|
@githubwalt for DeepStream 6.2, an opset of 16 should work well, but it’s recommended to test the default opset or consult the DeepStream documentation for a specific "opset - deepstream version" compatibility list. |
Beta Was this translation helpful? Give feedback.
-
|
Hi, I'm using the Jetson Orin Nano 8GB Developer Kit, and I followed the steps from the video and documentation to convert the YOLOv11 model to TensorRT. After building the model successfully, I encountered a segmentation fault (core dumped) error when trying to run the model. |
Beta Was this translation helpful? Give feedback.
-
|
I managed to fix the initial segmentation fault issue with the YOLOv11 model conversion to TensorRT. The model built successfully, and I was able to start it. However, now I’m encountering a new problem. After starting the pipeline, it crashed with the following message: **PERF: FPS 0 (Avg) ERROR from src_elem: Internal data stream error. ** ERROR: <parse_tiled_display:2013>: parse_tiled_display failed (deepstream-app:16369): GStreamer-CRITICAL **: 01:09:05.775: gst_element_send_event: assertion 'GST_IS_ELEMENT (element)' failed |
Beta Was this translation helpful? Give feedback.
-
|
@SashreekMallem it seems like there might be an issue with your DeepStream configuration files. Please ensure that the paths and parameters in |
Beta Was this translation helpful? Give feedback.
-
|
Will this guide work for every yolo model including YOLOv10 or it is exclusive for YOLOV11 alone. |
Beta Was this translation helpful? Give feedback.
-
|
@Blessing988 the guide is primarily designed for YOLOv11, but it may also be applicable to other YOLO models with some adjustments. For specific details, please refer to the documentation or test with your model. |
Beta Was this translation helpful? Give feedback.
-
|
do u have an implement of the bytetrack with deepstream? |
Beta Was this translation helpful? Give feedback.
-
|
@dao027 currently, we don't have a specific implementation of ByteTrack with DeepStream. However, you can refer to the DeepStream-Yolo repository for integrating YOLO models with DeepStream, which may assist in developing your own implementation. If you have further questions, feel free to ask! |
Beta Was this translation helpful? Give feedback.
-
|
|
Beta Was this translation helpful? Give feedback.
-
|
@yliuosu segmentation faults often indicate issues with memory or compatibility. Ensure your custom-trained model is properly exported to a compatible ONNX format with the correct input size and opset (e.g., 1280x1280 with |
Beta Was this translation helpful? Give feedback.
-
|
I found the reason. I need update the DeepStream-Yolo code to the latest version, otherwise the output of the network is different from the previous DeepStream-Yolo code. |
Beta Was this translation helpful? Give feedback.
-
|
Thank you for sharing your findings! Updating the code to match the latest version of |
Beta Was this translation helpful? Give feedback.
-
|
DeepStream 7.1 - Negotiation Errors & Best Practices for USB Webcam in Docker I'm trying to transition from example pipelines to a live USB webcam stream in DeepStream 7.1, but I'm struggling with negotiation errors when modifying configurations or building a custom Python pipeline. I've successfully run other pipelines, so I’m confident my hardware and passthroughs are working. However, integrating capture, display, and inference into a single working pipeline has been a challenge. Issues: Questions: I was thinking of using shared memory for a one to many model approach but im open to suggestions, I want to first get a USB webcam working as a known reference point. Any guidance would be greatly appreciated! |
Beta Was this translation helpful? Give feedback.
-
|
@aris-t for DeepStream USB webcam integration in Docker, ensure your pipeline uses |
Beta Was this translation helpful? Give feedback.
-
|
Thanks I got that far. The issue was in the NVIDIA deep stream auto builder. Once I got it worked out I was able to use the Gstreamer bindings to replicate it in C++. This is the source block: Source: |
Beta Was this translation helpful? Give feedback.
-
|
@aris-t great to hear you resolved the DeepStream auto-builder issue! For others facing similar challenges, the DeepStream configuration section in our Jetson guide shows how to structure camera input pipelines while maintaining GPU acceleration. Feel free to share your C++ implementation insights with the community if you'd like! 🚀 |
Beta Was this translation helpful? Give feedback.
-
|
So Glenn, |
Beta Was this translation helpful? Give feedback.
-
|
@aris-t thanks for reaching out! Currently, the DeepStream integration through the marcoslucianops/DeepStream-Yolo repository specifically supports YOLO detection models. For segmentation (masks) and pose (keypoints) tasks, you'd need to modify the custom parser and post-processing steps in the DeepStream pipeline, which isn't officially supported in our current implementation. We recommend monitoring the DeepStream-Yolo repo for potential updates or community contributions for multi-task support. For native Ultralytics deployment without DeepStream, our Python and CLI interfaces fully support all task types out-of-the-box via |
Beta Was this translation helpful? Give feedback.
-
|
Do we first have to set up our NVIDIA Jetson device with Ultralytics YOLO11 before following this guide? |
Beta Was this translation helpful? Give feedback.
-
|
@githubwalt yes, setting up your NVIDIA Jetson device with Ultralytics YOLO11 is a prerequisite. Please follow our Quick Start Guide: NVIDIA Jetson with Ultralytics YOLO11 before proceeding with this DeepStream deployment guide. |
Beta Was this translation helpful? Give feedback.
-
|
I mean, first do this https://docs.ultralytics.com/guides/nvidia-jetson/ |
Beta Was this translation helpful? Give feedback.
-
|
For details on temporary config file usage with DeepStream, refer to the DeepStream configuration for YOLO11 section in our NVIDIA Jetson deployment guide. This approach is commonly used in the community-driven DeepStream-Yolo repository for dynamic parameter adjustments without modifying original configs. |
Beta Was this translation helpful? Give feedback.
-
|
Sorry Glenn, but I can't find anything about details on temporary config file usage on the links you provided. |
Beta Was this translation helpful? Give feedback.
-
|
For temporary config file usage details specific to DeepStream-Yolo implementations, we recommend checking the community examples in the marcoslucianops/DeepStream-Yolo repository configuration files. This pattern is commonly used in production deployments but isn't explicitly documented in our core docs. Let us know if you'd like us to add an official example! |
Beta Was this translation helpful? Give feedback.
-
|
Thanks Glenn, no big deal. I know I read it somewhere in the past. Would be nice if you could add an official example for future reference. |
Beta Was this translation helpful? Give feedback.
-
|
@githubwalt thanks for the suggestion! We'll add a temporary config file example to our DeepStream deployment guide in a future update. For now, you can reference the existing configuration patterns shown in the NVIDIA Jetson with DeepStream guide. |
Beta Was this translation helpful? Give feedback.
-
|
Hi, I just followed the tutorial step by step and able to initiate the inference on the sample video file. However, all the detected objects are classified as "person" even the cars. I checked the "labels.txt" file and its path on the "config.txt" as well. I also set the "num-detected-classes" as 80 inside config.txt. My hardware is Jetson Xavier NX running Jetpack 5.0.2 and Deepstream 6.1 . I tried to omit "--simplify" option during conversion but it didn't help. Can you elaborate on what may be the reason for this misclassification ? |
Beta Was this translation helpful? Give feedback.
-
|
I changed the "labels.txt" file and deleted all the labels and set "dumb" as the only label. Everything is classified as "dumb" which means the cars are really classified as person by the weight file. So the onnx converted model is the issue from my perspective. I used yolo11m.pt and the suggested option flag during conversion. The only missing thing on my system is "tensorflowjs>=4.0.0" as it depends on flax and that depends on jaxlib which is not found by pip for any available installation candidate. "ERROR: Cannot install flax because these package versions have conflicting dependencies." So, is there any direct relation in between this pack and the conversion script to create onnx version of the weight file ? If there is, then how can I install this into my Xavier NX ? If there is not, what could be the problem for that strange behaving onnx file ? |
Beta Was this translation helpful? Give feedback.
-
|
So, I tried to generate onnx file with the yolo conversion command but then auto-converted engine file did not result in any detections on the frames. That's interesting. Then currently I'm suspicious that the engine file is defected. So I share the outputs of the deepstream-app command as below. Using winsys: x11 Building the TensorRT Engine WARNING: [TRT]: Tactic Device request: 307MB Available: 294MB. Device memory is insufficient to use tactic. WARNING: [TRT]: The getMaxBatchSize() function should not be used with an engine built from a network created with NetworkDefinitionCreationFlag::kEXPLICIT_BATCH flag. This function will always return 1. 0:07:11.696680172 2335644 0x70c7e30 INFO nvinfer gstnvinfer_impl.cpp:328:notifyLoadModelStatus:<primary_gie> [UID 1]: Load new model:/home/nvidia/DeepStream-Yolo/config_infer_primary_yoloV8.txt sucessfully |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for sharing the detailed logs. The warning about deserializing the engine file and subsequent rebuild suggests potential compatibility issues between your TensorRT version and the ONNX model. Let's address both classification and engine generation issues:
For immediate resolution: # Re-export with explicit opset and simplified ONNX
python3 export_yolo11.py -w yolo11m.pt --opset 17 --simplify --dynamicThe TensorFlowJS dependency isn't required for conversion. Focus on CUDA/cuDNN versions matching JetPack 5.0.2 requirements. For further debugging, refer to the NVIDIA Jetson with Ultralytics YOLO guide. |
Beta Was this translation helpful? Give feedback.
-
|
After checking the models structure of the generated onnx file, I realized that there is a transpose layer at the end; therefore the output is in float32[batch,8400,6] shape. However, when I execute the "yolo export model=yolo11n.pt format=onnx" command the generated onnx file has float32[1,84,8400] output. I did try to use that yolo11m.onnx instead of yolo11m.pt.onxx in the config file; however, there was no detection at all. So, I need to remove that transpose layer inside the yolo11m.pt.onnx to achieve float32[batch,8400,84] output format from my perspective. Still I'm confused that why this default pipeline is not working as expected based on the tutorial. |
Beta Was this translation helpful? Give feedback.
-
|
@gokcerSnmzck thanks for the detailed analysis. The output shape difference occurs because the DeepStream-Yolo repository requires a specific ONNX format that our For DeepStream deployments, please strictly follow the export process using the provided If you still face issues, please verify your DeepStream-Yolo implementation matches the latest version from their repository, as their parser logic dictates the required output format. |
Beta Was this translation helpful? Give feedback.
-
|
Do you have examples for models that are loaded on a Triton server? Also what if the source is a live source. Do you have gstreamer pipeline examples? |
Beta Was this translation helpful? Give feedback.
-
|
@Liagos for Triton Server deployment examples, see our Triton Inference Server guide. While we don't currently provide prebuilt GStreamer pipelines, our DeepStream integration demonstrates real-time stream handling through its GStreamer-based architecture - you can adapt the video source configuration for live inputs by modifying the |
Beta Was this translation helpful? Give feedback.
-
|
Hello. Can we use 2 YOLO11 models on 1 deepstream task? I'm running it on nvidia jetson orin nano but I couldn't seem to find the option to add another yolo11 model in the "deepstream_app_config.txt" file |
Beta Was this translation helpful? Give feedback.
-
|
You can alter the config.pbtxt for the model in question |
Beta Was this translation helpful? Give feedback.
-
|
@kalebmes Running two primary YOLO11 models within a single To achieve this, you would generally need to:
For guidance on building custom DeepStream applications, please refer to the official NVIDIA DeepStream documentation. |
Beta Was this translation helpful? Give feedback.
-
|
Hi, I've trained a custom classification model using yolo11n-cls.pt. However, when I try to export this model using export_yolo11.py, I encounter the following error: python3 export_yolo11.py -w /home/tns/Downloads/final-best-classifier-train1.pt Starting: /home/tns/Downloads/final-best-classifier-train1.pt It appears the error, AttributeError: 'tuple' object has no attribute 'transpose', occurs within the forward method of export_yolo11.py on line 31. This suggests that the input x at that point is a tuple, not a tensor, and therefore doesn't have the transpose method. Could you please advise on how to resolve this issue? |
Beta Was this translation helpful? Give feedback.
-
|
this step ->> error. anything I can do? |
Beta Was this translation helpful? Give feedback.
-
|
I'm currently working with the YOLOv11 Oriented Bounding Box (OBB) model exported in TensorRT format, and deploying it on an NVIDIA Jetson device (AGX Orin). While the model works in standard inference pipelines (.pt) , I’m facing challenges integrating it into a DeepStream-based application to process live RTSP video streams. There seems to be no existing reference implementation or sample Python code that demonstrates how to use the OBB variant of the model with DeepStream. Specifically, I'm looking for guidance on: 1- Integrating the YOLOv11 OBB TensorRT model with DeepStream (preferably using Python bindings). Any support, documentation, or examples from the team or community would be greatly appreciated. |
Beta Was this translation helpful? Give feedback.
-
|
Hello, i want to use the .engine model which is created by deepstream 7.1 in cpp or python. What should i do? Every instruction is just like prev versions! |
Beta Was this translation helpful? Give feedback.
-
|
@ARK1988 the @AbhisekRao unfortunately, YOLOv11 OBB (Oriented Bounding Box) models aren't currently supported by the DeepStream-Yolo repository, which focuses on standard detection tasks. Implementing OBB support would require developing custom parsers and post-processing logic for oriented bounding boxes within the DeepStream pipeline. We recommend using the standard YOLO11 detection models for DeepStream deployments. @bahare-mrt97 using TensorRT |
Beta Was this translation helpful? Give feedback.
-
|
Do you know how to get the detections to work within a ROI only in deepstream? |
Beta Was this translation helpful? Give feedback.
-
|
Hello, Deepstream Clarity question please. I notice the Nvidia Deepstream_python_app sample application use python file and one or more configuration text files. While Ultralytics Deepstream_Yolo uses two configuration text files.
Thank you for your help and time. |
Beta Was this translation helpful? Give feedback.
-
|
DeepStream-Yolo isn’t fundamentally “different” from For the YOLO26 ONNX export, it almost always comes down to a JetPack/DeepStream compatibility mismatch with |
Beta Was this translation helpful? Give feedback.
-
|
Hello. Thank you for the quick response. below is the terminal output. Starting: yolo26s.pt |
Beta Was this translation helpful? Give feedback.
-
|
That log shows PyTorch is actually exporting at opset 18 and then failing to down-convert to opset 17 ( python3 export_yolo26.py -w yolo26s.pt --opset 18If you’re on an older DeepStream/TensorRT stack that truly requires lower opset (e.g. DS 5.1), you’ll need the DS 5.1 settings from the guide ( |
Beta Was this translation helpful? Give feedback.
-
|
smh, User Error. :-) I forgot about the steps at https://docs.ultralytics.com/guides/nvidia-jetson/ . It works after following steps!!! One last issue for Segmentation ONNX conversion. Hopefully this is another user error. I got these error: Starting: yolo26s-seg.pt Thanks for your help and time. |
Beta Was this translation helpful? Give feedback.
-
|
That traceback is expected with |
Beta Was this translation helpful? Give feedback.
-
|
Title: Guidance Needed: Where to Modify Code to Add Custom Violation Logic and Modbus Output in DeepStream-YOLO Pipeline Hello DeepStream-YOLO Team, First of all, thank you for the excellent DeepStream-YOLO integration repository. I was able to successfully deploy a YOLO26 model using DeepStream on a Jetson Orin Nano (JetPack 6.1), and inference is working correctly using My Use CaseI am building an industrial safety system. So I need to integrate custom logic after detection results. My QuestionIn this DeepStream-YOLO setup, in which file or location should I add my custom logic? Specifically, I want to know:
Because I understand that:
So I want guidance on the recommended DeepStream architecture for this. My Current Setup
GoalAfter detection → evaluate violation → send Modbus 0/1 signal in real-time. I would really appreciate guidance on:
Thank you again for your great work 🙏 |
Beta Was this translation helpful? Give feedback.
-
|
You generally shouldn’t put violation/Modbus logic inside // Probe callback (attach to a pad after nvinfer/nvtracker, e.g. OSD sink pad)
static GstPadProbeReturn probe_cb(GstPad* pad, GstPadProbeInfo* info, gpointer user_data) {
GstBuffer* buf = (GstBuffer*)info->data;
NvDsBatchMeta* batch = gst_buffer_get_nvds_batch_meta(buf);
for (NvDsMetaList* lf = batch->frame_meta_list; lf; lf = lf->next) {
NvDsFrameMeta* f = (NvDsFrameMeta*)lf->data;
for (NvDsMetaList* lo = f->obj_meta_list; lo; lo = lo->next) {
NvDsObjectMeta* o = (NvDsObjectMeta*)lo->data;
// o->class_id, o->confidence, o->rect_params; compute “violation” then enqueue Modbus 0/1
}
}
return GST_PAD_PROBE_OK;
}If you share whether you need “violation” per-frame or latched with hysteresis (and whether you’re using |
Beta Was this translation helpful? Give feedback.
-
|
Hi @glenn-jocher — thanks, that makes sense. My requirements: Input: RTSP (H.264) on Jetson Orin Nano (JP 6.1 / DS 7.1) Throughput: ~25 FPS Tracking: Not using nvtracker (I only need a global safety signal, not per-person IDs) Decision: Per-frame violation, but I’m open to latching / hysteresis if that’s the recommended way to avoid flicker. Output: Modbus TCP to PLC; I need to continuously send 0/1 (currently planning to write a coil bit), e.g. modbus_write_bit(coil_addr, value) at ~25Hz (or “send on change” if better). Violation rule (current): I define a polygon zone (danger/no-go area). Violation = 1 if a detected person bbox is fully inside the polygon (all 4 corners inside). Violation = 0 otherwise. Questions to confirm best architecture: Probe location: since I’m not using tracker, is the best place a pad-probe on nvinfer src pad (or OSD sink pad)? Metadata fields: I’m reading from NvDsObjectMeta → class_id, confidence, rect_params (left/top/width/height). Is that the correct/standard way? Best practice for Modbus: should I publish every frame (25Hz), or only on state change with a small debounce/hysteresis window (e.g., require N consecutive frames to assert violation)? If you can point to the cleanest DeepStream sample to start from (e.g., deepstream-test3 or deepstream-app custom plugin approach) and the recommended probe pad in that pipeline, I’ll follow that pattern. Thanks again 🙏 |
Beta Was this translation helpful? Give feedback.
-
|
Hello |
Beta Was this translation helpful? Give feedback.
-
|
DeepStream already attaches detections/tracks as # inside a pad-probe callback
import pyds
batch_meta = pyds.gst_buffer_get_nvds_batch_meta(hash(gst_buffer))
l_frame = batch_meta.frame_meta_list
while l_frame:
frame_meta = pyds.NvDsFrameMeta.cast(l_frame.data)
l_obj = frame_meta.obj_meta_list
while l_obj:
obj = pyds.NvDsObjectMeta.cast(l_obj.data)
tid = int(obj.object_id) # tracker ID (requires nvtracker)
cid = int(obj.class_id) # class
r = obj.rect_params # bbox (left, top, width, height)
conf = float(obj.confidence)
# -> push to your local queue / file / socket for CrewAI
l_obj = l_obj.next
l_frame = l_frame.nextIf you reply with your JetPack/DeepStream versions and whether you’re currently running the stock |
Beta Was this translation helpful? Give feedback.
-
|
Thanks Glen. I am using Jetpack 6.2.1 and DeepStream 7.1. Deepstream-Yolo(deepstream_app_config.txt). How would the "deepstream_app_config.txt/config_infer_primary_yolo26" be mapped or linked to the deepstream python example you provided above? |
Beta Was this translation helpful? Give feedback.
-
|
# after you create the elements in your Python pipeline
pgie = Gst.ElementFactory.make("nvinfer", "primary-gie")
pgie.set_property("config-file-path", "config_infer_primary_yolo26.txt")
tracker = Gst.ElementFactory.make("nvtracker", "tracker")
# set tracker properties from your tracker config the same way (or load them via configparser) |
Beta Was this translation helpful? Give feedback.
-
|
Hello, Thank you for all your help to my issues. I have a usb camera with the below format and fails to run. How do I get deepstream to accept or convert the format below to the needed format for display. Deepstream 7.1/Jetpack 6.2.1deepstream app text file:[soruce0] v4l2-ctl -d /dev/video0 --list-formats-extioctl: VIDIOC_ENUM_FMT GST_DEBUG=3 deepstream-app -c deepstream_app_config_yolo26_1usbcam.txtSetting min object dimensions as 16x16 instead of 1x1 to support VIC compute mode. 0:00:00.401518311 6238 0xaaaad1ab46c0 INFO nvinfer gstnvinfer.cpp:684:gst_nvinfer_logger:<primary_gie> NvDsInferContext[UID 1]: Info from NvDsInferContextImpl::generateBackendContext() <nvdsinfer_context_impl.cpp:2195> [UID = 1]: Use deserialized engine model: /home/shamsee/DeepStream-Yolo/model_b1_gpu0_fp32.engine Runtime commands: NOTE: To expand a source in the 2D tiled display and view object details, left-click on the source. ** INFO: <bus_callback:291>: Pipeline ready ERROR from src_elem: Internal data stream error. ** INFO: <bus_callback:277>: Pipeline running ** INFO: <bus_callback:334>: Received EOS. Exiting ... Quitting |
Beta Was this translation helpful? Give feedback.
-
|
That [source0]
enable=1
type=1
camera-v4l2-dev-node=0
camera-width=80
camera-height=60
camera-fps-n=9
camera-fps-d=1
video-format=NV12If it still fails, first confirm GStreamer can negotiate your camera at |
Beta Was this translation helpful? Give feedback.
-
|
Thanks Glenn. Its working Sir!!!! |
Beta Was this translation helpful? Give feedback.
-
|
Below is the result from deepstream **PERF: FPS 0 (Avg) But i can get around 85 fps using ultralytics inference. why is that so. what is the need of using deepstream then. or i am doing anything wrong? |
Beta Was this translation helpful? Give feedback.
-
|
Those [osd]
enable=0
[tiled-display]
enable=0
[sink0]
type=1
sync=0If you paste your |
Beta Was this translation helpful? Give feedback.
-
|
@glenn-jocher Thank you so much for your reply. Now i understood and can see 120 Fps with this setting. But i have a doubt when do we need deepstream over normal ultralytics inferencing. Can you explain it |
Beta Was this translation helpful? Give feedback.
-
|
DeepStream is worth using when you need a production GStreamer pipeline around the model—hardware-accelerated decode/encode, zero-copy GPU memory, batching across multiple cameras/streams, RTSP ingest/egress, OSD/tiling, tracking, and easy composition with other DeepStream plugins—whereas “plain” Ultralytics inference is often the simplest/fastest path for single-stream or custom Python loops that just need model outputs; the DeepStream on Jetson guide and the TensorRT export docs are the best references, and feel free to drop your specific target pipeline (sources/sinks/stream count/latency needs) here so we can sanity-check the recommended setup. |
Beta Was this translation helpful? Give feedback.
and others
Uh oh!
There was an error while loading. Please reload this page.
-
guides/deepstream-nvidia-jetson/
Learn how to deploy Ultralytics YOLOv8 on NVIDIA Jetson devices using TensorRT and DeepStream SDK. Explore performance benchmarks and maximize AI capabilities.
https://docs.ultralytics.com/guides/deepstream-nvidia-jetson/
Beta Was this translation helpful? Give feedback.
All reactions