Practice Project 17

Computer Vision Application Project

Build an image processing pipeline using OpenCV. Apply Gaussian blurs, Canny edge detection, and thresholding, and run a Haar Cascade model for face detection.

Domain / Environment
Computer Vision / Conda VM
Difficulty
Intermediate (3/5)
Course Module
Computer Vision
Deliverables
Image processing script & face-detection output image
1. Image Processing & Detection Pipeline

The diagram below displays the computer vision pipeline. The input image goes through classical processing steps (blur, Canny edge detection, thresholding) to extract features, while the Haar Cascade classifier runs face detection to draw bounding boxes.

Input Image RGB Pixels Process 2. OpenCV Filters GaussianBlur / Canny Binary Threshold Detect 3. Haar Cascade haarcascade_xml detectMultiScale() Sliding window scan 4. Output canny_edges.png face_output.png Bounding boxes
2. Part 1: Step-by-Step Action Items & Key Execution Steps
STEP 1

Activate Python Virtual Environment

Point the terminal execution environment to the course conda sandbox environment.

$ conda activate ds_ai_ml
This targets active python libraries to the isolated virtual sandbox.
STEP 2

Create Project Folders inside Linux VM

Create a dedicated folder for the project files inside your guest VM home folder directory.

$ mkdir -p ~/Projects/opencv_app && cd ~/Projects/opencv_app
This sets up the working directory layout for the OpenCV code files.
STEP 3

Install OpenCV-Python via Pip

Install OpenCV libraries inside the active conda session.

$ pip install opencv-python numpy matplotlib requests
This installs the core OpenCV library, numpy arrays, and plotting tools.
STEP 4

Download Haar Cascade XML File

Download the pre-trained Haar Cascade face detection XML file from the OpenCV repository.

$ wget https://raw.githubusercontent.com/opencv/opencv/master/data/haarcascades/haarcascade_frontalface_default.xml
This downloads the XML file containing the features to run the face detection classifier.
STEP 5

Create image processing script file in VS Code

Launch VS Code and create the OpenCV pipeline script file.

Launch VS Code via terminal "code ." -> New File -> Type: cv_pipeline.py -> Paste Python code -> Save file
This registers the image processing operations and face detection loops in `cv_pipeline.py`.
STEP 6

Run and Verify the CV pipeline

Execute the script to run image processing operations and face detection.

$ python cv_pipeline.py
This runs Canny edge detection, thresholding, and face detection, saving the processed images to disk.
STEP 7

Open and verify generated image outputs

Open the output images using the default Linux desktop photo viewer to review the results.

$ xdg-open processed_filters.png && xdg-open face_detected.png
This command loads the image viewer application on the VM desktop to display the output images.
3. CV Execution Flow

The flowchart below outlines the computer vision execution flow. It details the steps from raw image loading and applying filters to running face detection and saving output images.

1. Load Image Read input image from workspace path cv2.imread() 2. Apply Filters Calculate Canny edges and thresholds cv2.Canny() 3. Load XML Load cascade XML detector weights CascadeClassifier() 4. Detect Faces Detect face bounds in input image detectMultiScale() 5. Save Image Write output images to workspace path cv2.imwrite()
4. Part 2: Complete Deliverable Assets & Production Templates

To run the computer vision pipeline, we need the Python script file. Below is a line-by-line explanation of the code, followed by the combined template.

Step-by-Step Code Construction

Lines 1 - 5

Import OpenCV and Numpy modules

Include OpenCV, Numpy arrays, and matplotlib plotting tools in the script.

import cv2 import numpy as np import matplotlib.pyplot as plt import os
These imports pull standard OpenCV image processing APIs, numpy matrices, and plotting libraries.
Lines 6 - 15

Generate Synthetic Image

Create a synthetic image with geometric shapes to apply CV filters on.

img = np.zeros((300, 300, 3), dtype=np.uint8) cv2.rectangle(img, (50, 50), (250, 250), (255, 255, 255), -1) cv2.circle(img, (150, 150), (50, 50, 50), (0, 0, 0), -1) cv2.imwrite("synthetic_input.png", img)
This generates a 300x300 pixel RGB image with a white square and black circle to test edge detection filters.
Lines 16 - 28

Apply Classical Image Filters

Apply Gaussian blurs, Canny edge detection, and binary thresholding to the generated image.

gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) blurred = cv2.GaussianBlur(gray, (5, 5), 0) edges = cv2.Canny(blurred, 50, 150) _, thresholded = cv2.threshold(blurred, 127, 255, cv2.THRESH_BINARY)
This converts the image to grayscale, applies a Gaussian blur, runs Canny edge detection, and applies binary thresholding.
Lines 29 - 42

Run Haar Cascade Face Detection

Load the XML classifier weights and run face detection on a dummy face matrix.

face_cascade = cv2.CascadeClassifier("haarcascade_frontalface_default.xml") dummy_face = np.zeros((200, 200, 3), dtype=np.uint8) # Draw circles representing eyes/mouth to simulate a face cv2.circle(dummy_face, (100, 100), (50), (255, 255, 255), -1) faces = face_cascade.detectMultiScale(dummy_face, 1.1, 4) for (x, y, w, h) in faces: cv2.rectangle(dummy_face, (x, y), (x+w, y+h), (0, 255, 0), 2)
This loads the Cascade XML file, runs face detection on a dummy face image, and draws bounding boxes around detected regions.

Production templates

1. Python script (Save as ~/Projects/opencv_app/cv_pipeline.py):

# cv_pipeline.py - Image filters and Haar Cascade face detection import cv2 import numpy as np import matplotlib.pyplot as plt import os def main(): print("=== Part 1: Generating Synthetic Image and Applying Filters ===") # Create 300x300 RGB image with white square and black circle img = np.zeros((300, 300, 3), dtype=np.uint8) cv2.rectangle(img, (50, 50), (250, 250), (255, 255, 255), -1) cv2.circle(img, (150, 150), 50, (0, 0, 0), -1) # 1. Convert to grayscale gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) # 2. Gaussian Blur blurred = cv2.GaussianBlur(gray, (5, 5), 0) # 3. Canny Edge Detection edges = cv2.Canny(blurred, 50, 150) # 4. Binary Thresholding _, thresh = cv2.threshold(blurred, 127, 255, cv2.THRESH_BINARY) # Save processing plots plt.style.use('dark_background') fig, axes = plt.subplots(1, 4, figsize=(15, 5)) axes[0].imshow(cv2.cvtColor(img, cv2.COLOR_BGR2RGB)) axes[0].set_title("Input Image") axes[1].imshow(gray, cmap="gray") axes[1].set_title("Grayscale") axes[2].imshow(edges, cmap="gray") axes[2].set_title("Canny Edges") axes[3].imshow(thresh, cmap="gray") axes[3].set_title("Thresholded") for ax in axes: ax.axis("off") plt.tight_layout() filters_filename = "processed_filters.png" plt.savefig(filters_filename, facecolor="#0f172a", edgecolor="none") print(f"Classical filters plot saved: {filters_filename}") print("\n=== Part 2: Running Haar Cascade Face Detection ===") cascade_path = "haarcascade_frontalface_default.xml" if not os.path.exists(cascade_path): print("Haar Cascade XML missing. Make sure to download it using wget in step 4.") return face_cascade = cv2.CascadeClassifier(cascade_path) # Create dummy face matrix (white circle on black background) dummy_face = np.zeros((300, 300, 3), dtype=np.uint8) cv2.circle(dummy_face, (150, 120), 60, (255, 255, 255), -1) # Head outline cv2.circle(dummy_face, (120, 100), 10, (0, 0, 0), -1) # Left Eye cv2.circle(dummy_face, (180, 100), 10, (0, 0, 0), -1) # Right Eye cv2.ellipse(dummy_face, (150, 160), (30, 10), 0, 0, 180, (0, 0, 0), -1) # Mouth gray_face = cv2.cvtColor(dummy_face, cv2.COLOR_BGR2GRAY) # Detect faces faces = face_cascade.detectMultiScale(gray_face, scaleFactor=1.05, minNeighbors=2, minSize=(50, 50)) print(f"Faces detected: {len(faces)}") for (x, y, w, h) in faces: cv2.rectangle(dummy_face, (x, y), (x+w, y+h), (0, 255, 0), 3) print(f" - Bounding Box: X={x}, Y={y}, Width={w}, Height={h}") face_output_filename = "face_detected.png" cv2.imwrite(face_output_filename, dummy_face) print(f"Face detection plot saved: {face_output_filename}") print("\n=== Computer Vision Project Successfully Complete! ===") if __name__ == "__main__": main()
5. Deliverables Summary

Verify that the following configurations and outputs exist inside your project workspace.

Created Files / Templates

  • ~/Projects/opencv_app/cv_pipeline.py - OpenCV pipeline script file.
  • ~/Projects/opencv_app/haarcascade_frontalface_default.xml - Pretrained Haar weights file.
  • ~/Projects/opencv_app/processed_filters.png - Saved classical filters plot.
  • ~/Projects/opencv_app/face_detected.png - Face detection bounding boxes plot.

Verification Artifacts / Execution Proof

  • Grayscale conversion, Gaussian blurring, Canny edge detection, and thresholding output plots saved to disk.
  • Haar Cascade classifier loaded successfully without path errors.
  • Correct bounding boxes drawn around detected regions in output images.
6. Closing Explanation: Why We Did This & What It Accomplishes

Architectural Intent & Operational Impact

Why We Did This

What This Accomplishes

Data Science & AI/ML Training Program | Practice Project 17 | Computer Vision