Please provide the following information when requesting support.
• Hardware Jetson TX1
• Jetpack 4.4
• TensorRT 7.X
• Network Type (Yolo_v4) Object Detection
• TLT Version (3.21.08)
• Training spec file(If have, please share here)
• I want to inference of tao generated model in python without using deepstream is it possible
It is possible. You can directly deploy the etlt model with triton-app.
GitHub - NVIDIA-AI-IOT/tao-toolkit-triton-apps: Sample app code for deploying TAO Toolkit trained models to Triton
You can also refer to the preprocessing and postprocessing in
tao-toolkit-triton-apps/tao_client.py at main · NVIDIA-AI-IOT/tao-toolkit-triton-apps · GitHub and
tao-toolkit-triton-apps/yolov3_postprocessor.py at main · NVIDIA-AI-IOT/tao-toolkit-triton-apps · GitHub
More, you can also search and refer to some topics in forum.
Actually my model input size is 1472X960. So if resize the image without changing aspect ratio the resized image size is 1472X828. then how can i feed this image to inference.
Here it is running trtexec like this:
/usr/src/tensorrt/bin/trtexec --loadEngine=trt.engine --verbose
&&&& RUNNING TensorRT.trtexec [TensorRT v8001] # /usr/src/tensorrt/bin/trtexec --loadEngine=trt.engine --verbose
[11/01/2021-09:15:32] [I] === Model Options ===
[11/01/2021-09:15:32] [I] Format: *
[11/01/2021-09:15:32] [I] Model:
[11/01/2021-09:15:32] [I] Output:
[11/01/2021-09:15:32] [I] === Build Options ===
[11/01/2021-09:15:32] [I] Max batch: 1
[11/01/2021-09:15:32] [I] Works…
Bounding box indices[ [x1,y1,x2,y2],[…]…] of the detections.
Here is the full script (it’s quite basic):
import tensorrt as trt
TRT_LOGGER = trt.Logger(trt.Logger.VERBOSE)
DTYPE_TRT = trt.float32
import pycuda.driver as cuda
from PIL import Image
import numpy as np
path_img = "image.jpg"
offsets = ( 103.939, 116.779, 123.68 )
yolo_reso = (3, 768, 1024)
# Simple helper data class that's a little nicer to use than a 2-tuple
# from TRT Python sample code
Ok, with your recomendations i found a workinkg example of inferring with yolo v4. But i still have some issues:
The model im using is custom YOLO v4 trained with our own dataset. With the example of TLT (tlt_vc_samples_v1.1.0/yolo_v4/yolo_v4.ipynb). It is trained for Person, Car and Two_wheels.
trt2-yolo.engine - Google Drive
For exporting the model we use:
!tlt yolo_v4 export -m $USER_EXPERIMENT_DIR/experiment_dir_unpruned/weights/yolov4_resnet18_epoch_080.tlt \
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.