mctthermalface 0.2.2

No project description provided

Face Detection on Thermal Image

Installation

Before installing, make sure the following packages are available:

• Numpy
• PyTorch (I recommend to install your PyTorch from following website)
• PIL
• OpenCV

To Install, run the following command:

pip install mctthermalface

Only available in PIP, not in Conda

Usage

For 3 channels image (8-bit)

import mctthermalface
from PIL import Image

detector = mctthermalface.Detector('thermal_face_detection.pt')
img = Image.open("img_path.png")
bbox = detector.detect(img)

Fpr 1 channels image (16-bit) stored in csv

import mctthermalface
import cv2
import numpy as np

detector = mctthermalface.Detector('thermal_face_detection.pt')

img_1channel = np.loadtxt("img_path.csv", delimiter=",")
img_1channel = cv2.normalize(img_1channel, None, 0, 255, cv2.NORM_MINMAX, cv2.CV_8U)
bbox = detector.detect(img_1channel)

Bounding Box Format

bbox = [xmax, xmin, ymax, ymin]

Model Download

Click Here To Download thermal_face_detection.pt

Sample Usage

Check this notebook

Credits

Original Works by: Diza Febriyan Hasal on his Bachelor Thesis about Detecting Face in Thermal Image. Reworked by Martin Clinton Manullang to adjust the performance, adaptability, and compatibility (2023).