Build Your own Self Driving Car [Course -2]

Train a stop sign detector using a Haar cascade model by preparing stop sign images as positive samples. Print and arrange the images for camera-based training while adjusting for orientation.

Learn to write a C++ program to capture and save grayscale images of a stop sign using a Raspberry Pi camera, creating 40 positive samples for neural network training.

Capture positive samples of a stop sign with the car camera from multiple angles. Organize them into a positive stop sign folder and prepare for subsequent negative samples.

Captures negative samples greater than positives, using a Raspberry Pi to collect about 300 images from various angles, organizes positives and negatives, and prepares for training.

Download and install the tool, load the positive and negative folders, crop stop-sign images, and save the cropped positives and negatives for neural network training.

Train a Haar cascade stop sign detector using positive and 300 negative samples, 20 stages, and 32 by 32 image windows to generate the stop cascade classifier, then test.

Load the stop cascade xml into a C++ program on a Raspberry Pi, initialize a cascade classifier, and implement a stop sign detection workflow with error checks.

Process each frame to detect a stop sign with a cascade classifier in OpenCV, convert to grayscale, equalize histogram, draw a red rectangle around detected signs, and display the result.

Test the self driving car project by placing a stop sign on the track, running the program, and confirming stop sign detection after tweaking the color space to rgb.

Compute the real distance from a stop sign to a wall by cropping frames, measuring pixel differences, and solving two linear equations to estimate distance in centimeters.

Solve two linear equations to determine m and c, then apply them in a C++ program to compute and display distance in centimeters.

Demonstrate C++ program on a Raspberry Pi that estimates stop sign distance via linear regression and stops the robot car within safe range using if statements and go to label.

Learn to program an Arduino Uno in C++ to control motors and detect a stop sign, using if conditions, delays, and serial data to drive forward or stop.

Final testing demonstrates stop sign detection at 60 to 70 centimeters with a Raspberry Pi and Arduino, resetting distance and stop variables for subsequent object detection.

Capture 40 positive samples of the object using the sampling program from multiple angles around a dummy car, then organize images into a dedicated positive folder for training.

Select positive samples for object from car images by cropping the front or back portions and save them to a designated output folder.

Train a cascade for object detection by preparing positive and negative samples, configuring 20–25 stages, and using 42 by 32 samples, then test the cascade on images to detect objects.

Develop object detection for a self-driving car by creating an object cascade, adapting the stop sign detection function, and displaying results in named windows on the Raspberry Pi.

trim the frame to define a region of interest for object detection, then use linear equations to map pixel distances to real-world distance.

Write a c++ arduino program to respond to object detection from a raspberry pi by stopping, turning, and moving forward with delays. Compile and test the self-driving car.

Implement lane change by tracking detections with two counters, i and g; increment i on detection, trigger the lane switch via a dedicated function, and tune g to 10.

Final testing for object detection tunes lane-change timing and values j, gso, and gdp on a raspberry freeway, with object training and detection completed and traffic light detection upcoming.

Build a traffic light model with LCD indicators powered by double-A batteries, each controlled by a switch, and use a white border around the green light to improve cascade training.

Capture 60 positive red-light samples at multiple angles and brightness to train a self-driving car detector, then organize images into a dedicated positive images folder.

Capture negative samples for red and green traffic lights, copy and paste into the end folder, and run to collect 30 green-light images, totaling 330 negatives before capturing positive samples.

Prepare training data by loading positive traffic light images, cropping them, and organizing them into an output folder, then train using both positive and negative samples.

Train a cascade model for traffic lights using a labeled dataset of positive and negative images, adjust stages, and test red and green light detection for robust red-light recognition.

Load the xml cascade for traffic light detection, copy and rename a prior function, instantiate cascade and related variables, integrate into the main loop, and display the resulting frames.

Calibrate a traffic-light detector for a self-driving car using linear equations and pixel-distance measurements, retraining beyond 15 cm and tuning sample width, height, and aspect ratio.

Solve two linear equations via linear regression to estimate distance, determine M and C, and test distance and traffic light detection in a self-driving car program.

Practice final testing of an Arduino-based self-driving car, wiring, compiling, and uploading code to run traffic light and stop sign detection, with training images to refine the detector.