Jetson Inference - Python - DetectNet - Send mail

Marco972 · November 1, 2019, 2:52am

Hi,

I used jetson-inference with detectnet-camera.py, how to save a image (file captured) when a motion detected ?

Thanks for your reply

Marco

SunilJB · November 11, 2019, 4:46pm

Hi,

One of the way is to use opencv for the motion detection using the custom logic to check diff between the frames.
Also, jetcam has some in-build functions to attach callbacks to the camera value changes. Please check below link if you can use that in your solution.

github.com

NVIDIA-AI-IOT/jetcam/blob/master/notebooks/usb_camera/usb_camera.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First, we create our camera class like this.  Please note, you can only create one USBCamera instance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from jetcam.usb_camera import USBCamera\n",
    "\n",
    "camera = USBCamera(width=224, height=224, capture_width=640, capture_height=480, capture_device=1)"
   ]
  },

This file has been truncated. show original

For image capture you can use “cv2.imencode” to encode current frame values in jpg/png format and then write the encoded values to file in binary mode using “open(‘*.png’, ‘wb’)”.

Please let us know in case of queries related to TRT.

Thanks