Can running classify.py additionally capture the original image at the camera’s native size?

Lin.Hsu · January 9, 2023, 11:09am

Question/Issue:

When using running classify.py, the image will be compressed to the size of the model for identification. Is it possible to capture the original image of the camera’s native size? In this way, I can also collect data and return it to add new training images.

For example: the camera resolution is 640*480, I want to save the original image every five seconds, thank you for your reply!

OmarShrit · January 9, 2023, 11:23am

Hi Lin,

I would probably use classify-image script:

github.com

edgeimpulse/linux-sdk-python/blob/03e6026438bc7f03754c872acc6567dd5734ce87/examples/image/classify-image.py

#!/usr/bin/env python

import device_patches       # Device specific patches for Jetson Nano (needs to be before importing cv2)

import cv2
import os
import sys, getopt
import numpy as np
from edge_impulse_linux.image import ImageImpulseRunner

runner = None

def help():
    print('python classify-image.py <path_to_model.eim> <path_to_image.jpg>')

def main(argv):
    try:
        opts, args = getopt.getopt(argv, "h", ["--help"])
    except getopt.GetoptError:
        help()

This file has been truncated. show original

Inside there is get_features_from_image function that you can use to recover the original frame.

Best regards,
Omar

Lin.Hsu · January 11, 2023, 9:15am

Hi Omar,

I understand, thank you for your reply.

Lin