Building A Face Detection App with Streamlit and OpenCV

In this tutorial we will be building a simple face detection app with streamlit  and OpenCV. We will be using streamlit to help us with our UI. This is going to be a simple Computer Vision App using OpenCV in python. This is how our app will look like after we are done building it.

Installation

Note: For our drag and drop/ file upload  section we will need to use the newer versions of Streamlit precisely from version 0.53 and upwards.

pip install streamlit opencv-python

Let us check the basic structure and workflow for our app.

Our app will have two main sections, one for manipulating our uploaded image and another for detecting certain features – such as face,smiles,eyes in our uploaded image.

 

Image Manipulation

For our image manipulation we will be using PILLOW’s features to help us with the contrast as well as the brightness.

• Contrast
• Brightness
• Blurring

Face Detection

For our face detection we will be using the haarcascade xml files from opencv here

• Detect Face
• Detect Eyes
• Detect Smiles
• Cartonize
• Cannize

Our app will have a drag and drop/file upload section for uploading our images using the st.file_uploader function of Streamlit.

The image uploaded will be opened with python pillow package as well as with numpy to convert them into numpy arrays for the next section.

We will then create individual function to process and manipulate our images using pillow and opencv.

You can check the entire code below.

# Core Pkgs
import streamlit as st 
import cv2
from PIL import Image,ImageEnhance
import numpy as np 
import os


face_cascade = cv2.CascadeClassifier('frecog/haarcascade_frontalface_default.xml')
eye_cascade = cv2.CascadeClassifier('frecog/haarcascade_eye.xml')
smile_cascade = cv2.CascadeClassifier('frecog/haarcascade_smile.xml')

def detect_faces(our_image):
	new_img = np.array(our_image.convert('RGB'))
	img = cv2.cvtColor(new_img,1)
	gray = cv2.cvtColor(new_img, cv2.COLOR_BGR2GRAY)
	# Detect faces
	faces = face_cascade.detectMultiScale(gray, 1.1, 4)
	# Draw rectangle around the faces
	for (x, y, w, h) in faces:
				 cv2.rectangle(img, (x, y), (x+w, y+h), (255, 0, 0), 2)
	return img,faces 


def detect_eyes(our_image):
	new_img = np.array(our_image.convert('RGB'))
	img = cv2.cvtColor(new_img,1)
	gray = cv2.cvtColor(new_img, cv2.COLOR_BGR2GRAY)
	eyes = eye_cascade.detectMultiScale(gray, 1.3, 5)
	for (ex,ey,ew,eh) in eyes:
	        cv2.rectangle(img,(ex,ey),(ex+ew,ey+eh),(0,255,0),2)
	return img

def detect_smiles(our_image):
	new_img = np.array(our_image.convert('RGB'))
	img = cv2.cvtColor(new_img,1)
	gray = cv2.cvtColor(new_img, cv2.COLOR_BGR2GRAY)
	# Detect Smiles
	smiles = smile_cascade.detectMultiScale(gray, 1.1, 4)
	# Draw rectangle around the Smiles
	for (x, y, w, h) in smiles:
	    cv2.rectangle(img, (x, y), (x+w, y+h), (255, 0, 0), 2)
	return img

def cartonize_image(our_image):
	new_img = np.array(our_image.convert('RGB'))
	img = cv2.cvtColor(new_img,1)
	gray = cv2.cvtColor(new_img, cv2.COLOR_BGR2GRAY)
	# Edges
	gray = cv2.medianBlur(gray, 5)
	edges = cv2.adaptiveThreshold(gray, 255, cv2.ADAPTIVE_THRESH_MEAN_C, cv2.THRESH_BINARY, 9, 9)
	#Color
	color = cv2.bilateralFilter(img, 9, 300, 300)
	#Cartoon
	cartoon = cv2.bitwise_and(color, color, mask=edges)

	return cartoon


def cannize_image(our_image):
	new_img = np.array(our_image.convert('RGB'))
	img = cv2.cvtColor(new_img,1)
	img = cv2.GaussianBlur(img, (11, 11), 0)
	canny = cv2.Canny(img, 100, 150)
	return canny

def main():
	"""Face Detection App"""

	st.title("Face Detection App")
	st.text("Build with Streamlit and OpenCV")

	activities = ["Detection","About"]
	choice = st.sidebar.selectbox("Select Activty",activities)

	if choice == 'Detection':
		st.subheader("Face Detection")

		image_file = st.file_uploader("Upload Image",type=['jpg','png','jpeg'])

		if image_file is not None:
			our_image = Image.open(image_file)
			st.text("Original Image")
			# st.write(type(our_image))
			st.image(our_image)

		enhance_type = st.sidebar.radio("Enhance Type",["Original","Gray-Scale","Contrast","Brightness","Blurring"])
		if enhance_type == 'Gray-Scale':
			new_img = np.array(our_image.convert('RGB'))
			img = cv2.cvtColor(new_img,1)
			gray = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
			# st.write(new_img)
			st.image(gray)
		elif enhance_type == 'Contrast':
			c_rate = st.sidebar.slider("Contrast",0.5,3.5)
			enhancer = ImageEnhance.Contrast(our_image)
			img_output = enhancer.enhance(c_rate)
			st.image(img_output)

		elif enhance_type == 'Brightness':
			c_rate = st.sidebar.slider("Brightness",0.5,3.5)
			enhancer = ImageEnhance.Brightness(our_image)
			img_output = enhancer.enhance(c_rate)
			st.image(img_output)

		elif enhance_type == 'Blurring':
			new_img = np.array(our_image.convert('RGB'))
			blur_rate = st.sidebar.slider("Brightness",0.5,3.5)
			img = cv2.cvtColor(new_img,1)
			blur_img = cv2.GaussianBlur(img,(11,11),blur_rate)
			st.image(blur_img)
           
		elif enhance_type == 'Original':
			st.image(our_image,width=300)
		else:
			st.image(our_image,width=300)



		# Face Detection
		task = ["Faces","Smiles","Eyes","Cannize","Cartonize"]
		feature_choice = st.sidebar.selectbox("Find Features",task)
		if st.button("Process"):

			if feature_choice == 'Faces':
				result_img,result_faces = detect_faces(our_image)
				st.image(result_img)

				st.success("Found {} faces".format(len(result_faces)))
			elif feature_choice == 'Smiles':
				result_img = detect_smiles(our_image)
				st.image(result_img)


			elif feature_choice == 'Eyes':
				result_img = detect_eyes(our_image)
				st.image(result_img)

			elif feature_choice == 'Cartonize':
				result_img = cartonize_image(our_image)
				st.image(result_img)

			elif feature_choice == 'Cannize':
				result_canny = cannize_image(our_image)
				st.image(result_canny)




	elif choice == 'About':
		st.subheader("About Face Detection App")
		st.markdown("Built with Streamlit by [JCharisTech](https://www.jcharistech.com/)")
		st.text("Jesse E.Agbe(JCharis)")
		st.success("Jesus Saves @JCharisTech")



if __name__ == '__main__':
		main()	

With Streamlit you can build awesome apps  within a short time that is production ready. You can get more on building machine learning web apps in this mega course or on Udemy.

Check out the full video tutorial on our channel here

Thanks For watching

Jesus Saves

By Jesse E.Agbe(JCharis)