Pedestrian Detection for Node

Uses opencv4nodejs to do pedestrian detection. Detector is already configured with default parameters so you can just call detect and pass in the image you want detected as a buffer array. If you know what you are doing you can change the parameters of detect by passing parameters into optionalInit once before calling detect.

See test1.js and test2.js as examples.

test1.js uses images from the images folder and sends each one through the detect algorithm and saves the results in the results folder.

test2.js get an image from a camera and send that image through the detect algorithm and saves the results in the results folder. test2.js gets its camera configuration from cam-config.json. Copy file cam-config-placeholder.json, rename it to cam-config.json and put your camera details in it.

config

This module depends on the module opencv4nodejs. Add the opencv4nodejs configuration to your package.json:

}
  "name": "your app",
  ...
  "opencv4nodejs": {
    "autoBuildFlags": "-DBUILD_LIST=core,highgui,imgcodecs,objdetect,video,videoio",
    "autoBuildOpencvVersion": "4.3.0"
  }
}

This configuration tells opencv4nodejs to only include these openCV modules. By default it installs much more. It also tells opencv4nodejs to install openCV version 4.3.0.

If the opencv4nodejs auto build script does not work, perhaps a manual install of openCV would work better. See instructions here.

passing options

Pass options to optionalInit function once before calling detect.

const fullyLoadedOptions = {
  rectColor: { // red, green, blue
    r: 255,
    g: 255,
    b: 0
  },
  rectLineThickness: 2, // pixels,
  rectLineType: require('opencv4nodejs').LINE_8,
  outputFileType: '.jpg',
  hitThreshold: 0,
  winStride: {
    width: 4,
    height: 4
  },
  padding: {
    width: 8,
    height: 8
  },
  scale: 1.05,
  overlapThresh: .65,
};

pedestrianDetect.optionalInit(fullyLoadedOptions);
...
const result = await pedestrianDetect.detect(image);
...

option explanation

hitThreshold: Threshold for the distance between features and SVM classifying plane. Helpful for extremely high rate of false-positive detections. winStride: How quickly searching window parses an image. Lower number means high CPU intensive and more accurate. Higher number means less CPU intensive but less accurate. Min value of 1. padding: the number of pixels in both the x and y direction in which the sliding window ROI is “padded” prior to HOG feature extraction. Typical values for padding include (8, 8), (16, 16), (24, 24), and (32, 32) scale: Controls the resize factor while process is repeatedly resizing image to find pedestrians. Smaller results in more resizes, more accurate and more CPU usage. Lower is less resizes, less accurate and less CPU usage. Minimum is 1.01. overlapThresh: Used for non-maximum suppression, corrects for multiple detection of the same pedestrian. Amount of overlap per detection before counting it as a single detection.

See test2.js for a simple example of passing options.

results

The detect function returns a promise that resolves to an object

{
  img: <image>, // the image, with rectangles of anything detected.
  time: <number>, // milliseconds of how long the detection took. Useful in tuning performance.
  found: <number> // number of pedestrians detected. Matches the number of rectangles on img.
}

rectangle only mode

Outputs a png image that is the same dimensions as input image and is entirely transparent except for where rectangles are drawn.

const rectangleOnlyOptions = {
  rectanglesOnly: true,
  rectanglesOnlyBorder: [0, 255, 255] // when bitwiseNot, this is blue
};
pedestrianDetect.optionalInit(rectangleOnlyOptions);

other

Inspired by https://www.pyimagesearch.com/2015/11/09/pedestrian-detection-opencv/ pedestrian-detect.js is based on pedestrian-detect.py