Tag Archives: Fusion AI Lab Kit

AI On the Edge, Raspberry Pi

AI on the Edge LESSON 31: Facial Recognition in OpenCV Using Haarcascades

Hey everyone, this is Paul McWhorter from toptechboy.com. Welcome back to our AI on the Edge tutorial series. If you’ve made it this short distance through the course, you are doing fantastic.

Today, we are stepping away from simply pulling a clean, high-frame-rate video stream off our hardware, and we are finally going to start doing some real Computer Vision. We are going to teach our machine how to look at an image, process it, and actually locate human faces in real time.

Go ahead and grab yourself a hot cup of coffee. Turn off your cell phone, close your other browser tabs, and let’s get ready to write some serious engineering code.

The Core Concept: What is a Haar Cascade?

Up until now, we’ve been focused on the plumbing—getting the camera configured, setting up the dimensions, and flipping the frames so they look right on our screens. Today, we introduce the Haar Cascade Classifier.

Think of a Haar Cascade as an incredibly smart, pre-trained statistical filter. Decades ago, researchers realized that human faces share universal geometric patterns of light and dark. For example:

  • The bridge of your nose is almost always brighter than the shadows on either side of it.

  • Your eye sockets are consistently darker than your forehead or your cheekbones.

OpenCV comes pre-packaged with these mathematical descriptions of a face. The algorithm works by taking a tiny “sliding window” and dragging it across your video frame pixel by pixel, looking for those specific arrangements of light and dark gradients. When it finds a cluster that matches the math, it flags it as a face.

Why Grayscale Matters in Machine Learning

If you look closely at our pipeline logic for today, the absolute first thing we do after capturing a raw frame from our camera is strip away all the color and convert the image to grayscale.

Why do we do this? Color is a computational luxury we cannot afford when doing real-time edge processing. To a computer, a color image consists of three separate channels: Red, Green, and Blue. That means for a standard resolution, the processor has to crunch three times the data.

Because Haar Cascades only care about the contrast—the relationship between light and dark areas—the actual color of your skin, your hair, or your shirt is completely irrelevant. By dropping the frame down to a single grayscale intensity channel, we cut our processor’s workload by 66% instantly. This is how we keep our edge hardware running lightning-fast without lagging the video feed.

Tuning the Detection Engine

When we tell OpenCV to look for faces using the detectMultiScale function, we pass three critical parameters that you need to master. If you don’t tune these right, your program will either miss faces entirely or start thinking your coffee mug or a pattern on the wall is a human being.

  • scaleFactor=1.1: A single Haar Cascade model is trained at a fixed size. But a face could be right next to the lens (huge) or all the way across the room (tiny). The scaleFactor tells OpenCV to shrink the image by 10% on each pass, creating a “layer cake” of images at different scales so the sliding window can catch faces of any size.

  • minNeighbors=5: As the sliding window moves, it might find dozens of potential matches around your eyes and nose. This parameter states that a face must be detected in at least 5 overlapping bounding boxes before the system officially declares, “Yes, that is a human face.” Raising this number reduces false positives but makes the system less sensitive.

  • minSize=(60,60): This tells the algorithm to completely ignore any detected objects smaller than a 60×60 pixel block. This prevents the system from wasting CPU cycles trying to analyze tiny bits of background noise in the distance.

The Secret to a Smooth FPS Counter

You will notice a very specific mathematical formula used to calculate our Frames Per Second (FPS) in this lesson. If you simply calculate $1 / \Delta T$ every single frame, your FPS display will jitter wildly on screen, flashing unreadable numbers back and forth because individual frames might take a millisecond more or less to process.

To fix this, we implement a digital Low-Pass Filter.

Every loop, we take 95% of our previous FPS value and add only 5% of our current instantaneous speed measurement. This creates a beautiful, smooth running average that responds immediately to system slowdowns but remains steady and completely legible on the screen. It is an elegant engineering solution to a common UI headache.

Homework Assignment!

You didn’t think you were going to get out of here without homework, did you? No shortcuts in this classroom!

Now that your program can successfully locate faces and draw a bounding box around them, it’s time to take it to the next level. Your assignment is to modify this program to isolate and track only the single largest face in the frame.

If multiple people walk into the camera’s view, your script must evaluate the dimensions of the returned bounding boxes, determine which person is closest to the lens (the largest box), and draw a bright green rectangle around only that leader face, while ignoring everyone else.

Here is the code developed in today’s lesson.

 

AI On the Edge, Raspberry Pi

AI on the Edge LESSON 29: Improved Proportional Object Tracking with Pan Tilt Camera

AI on the Edge LESSON 29: Improved Proportional Object Tracking with Pan Tilt Camera

Hey everyone, Paul McWhorter here from TopTechBoy.com. Welcome back to our channel, where we learn to build real, intelligent systems on edge hardware. Go ahead and grab yourself a nice hot cup of coffee or a big glass of iced tea, because today we are going to completely revolutionize the way our robotic pan-tilt camera interacts with the physical world.

In Lesson 28, we successfully closed the loop. We got our camera to physically move and track an object using the error signal calculated from our OpenCV bounding box. It worked, but let’s be honest with ourselves: it was clunky. It was a crude, incremental system that moved the camera by exactly one lazy degree at a time, regardless of whether the target was right next to the crosshairs or flying across the room. It was jerky, it hunted back and forth, and it just wasn’t elegant old-school engineering.

Today, we are throwing away that clunky incremental logic and replacing it with something beautiful: Proportional Control.

The Problem with Lazy Incremental Steps

Before we fix our control loop, we need to understand exactly why our previous system struggled. In our last script, we used conditional statements to see if the error was positive or negative, and then adjusted our angles by a fixed step of 1 or -1.

This created two major engineering flaws:

  • Lagging on Large Errors: If you suddenly jerked the object 400 pixels away from the center, the camera would take forever to catch up because it could only step at a constant speed of one degree per loop iteration.

  • Hunting and Jitter on Small Errors: When the object finally got close to the center, the camera would overshoot by a full degree, trip the opposite condition, and step back. It would constantly “hunt” back and forth across the target, buzzing your hardware to pieces.

The Elegance of Proportional Control

In real-world automation, we don’t use rigid, conditional step-programming to move hardware. We use mathematics. We want the camera’s reaction to be completely proportional to the size of the mistake it is trying to correct.

If the object is a massive distance away from the center crosshairs, we want the servo to take a massive, aggressive leap to catch up instantly. As the object gets closer and closer to the center, we want the camera to automatically slow down and gently glide into place. When the error drops to zero, the physical adjustment should naturally drop to zero.

The magic of this approach is that it allows us to completely eliminate the bulky conditional statements and artificial deadbands we wrote last time. The algebra naturally handles the direction and magnitude of the movement.

Breaking Down the Math and Logic

To achieve this fluid motion, we take our raw error signal—the distance in pixels between our frame center and the object center—and apply a scaling factor, known in control theory as Gain.

In this updated system design, we take our error and divide it down. Specifically, we divide the pixel error by 50, and then split that in half by dividing by 2. Mathematically, this means we are scaling our pixel error down by a factor of 100.

  • If your object is 300 pixels off-center, the math calculates an instantaneous adjustment of 3 degrees, quickly snapping the camera toward the target.

  • If the object is only 10 pixels off-center, the adjustment becomes a tiny fraction of a degree (0.1), smoothly stabilizing the camera track.

Precision Tracking with Floating-Point Variables

Because we are dividing our pixel error down by 100, our angular adjustments will almost always be fractional decimals rather than clean integers. If we tried to store these angles as standard integers, our program would truncate those decimals, completely throwing away our precise micro-adjustments and causing the camera to stall out.

To make this system work perfectly, we maintain our accumulation variables as high-precision floating-point numbers. The script constantly adds and subtracts these fractional updates over time behind the scenes. We only cast the final calculated angle to a clean, rounded integer at the absolute last microsecond right as we pass the position command to the physical servo motors.

Visual Tuning and Smooth Performance

You will notice a massive visual upgrade when running this refined loop. To match our new high-precision math, we tighten up our tracking reticle overlay, shrinking our target circle down from a radius of 40 to a crisp 30 pixels. We also change our dynamic bounding box to a bright, vibrant yellow to make our tracking visually pop on screen.

When you fire up this loop and wave your object around, you will see a night-and-day difference compared to last week. The lazy, robotic stutter is completely gone. The pan-tilt mount tracks with an organic, fluid motion, actively accelerating and decelerating to mirror your movements perfectly.

 

AI On the Edge, Python

AI on the Edge LESSON 28: Use Pan Tilt Camera to Track Object of Interest in OpenCV

Hey everyone, Paul McWhorter here from TopTechBoy.com. Welcome back to our channel, where we don’t just write abstract software—we build real, physical, intelligent machines. Go ahead and grab yourself a nice hot cup of coffee or a big glass of iced tea, because today we are closing the loop between the digital world of computer vision and the physical world of robotics.

In our last lesson, we successfully taught OpenCV how to find a specific color, isolate the largest shape, and draw a beautiful green bounding box around it. That was great, but it had a massive limitation: if your object moved off the edge of the frame, it was gone forever. The camera just sat there, blind and helpless.

Today, we change that. We are taking that tracking data from our code and using it to command a physical pan-tilt mechanism powered by two servos on our Fusion Hat. By the time we finish today, your camera will physically turn, tilt, and hunt down your object, keeping it locked dead center in the middle of your video feed.

The Big Leap: Closing the Loop

What we are building today is a foundational concept in automation engineering known as a feedback control loop.

Up until now, your camera was an open-loop observer. It saw things, but it couldn’t react physically. To make an autonomous tracking system, we need to implement a simple pipeline:

  1. Sense: The camera captures the image frame.

  2. Think: OpenCV finds the target object and calculates its position.

  3. Act: The script commands the hardware servos to move the camera mount to correct any positioning errors.

The Mathematics of the Target Error

To make a camera track an object, we have to define what “perfect tracking” looks like to a computer. Perfect tracking means the center of our tracked object is sitting exactly at the center of our video frame.

Because we are running our camera at a crisp resolution of 1280×720, the mathematical center of our universe is fixed. We calculate our frame’s horizontal and vertical centers by dividing our dimensions in half. This gives us a permanent anchor point right in the middle of our grid.

When an object appears on screen, our contour detection gives us its bounding box. We calculate the exact center of that box by taking its starting coordinate and adding half of its width and height. Now we have two sets of coordinates:

  • Where we want the object to be (The Frame Center).

  • Where the object actually is (The Box Center).

The difference between where the object is and where it belongs is called the Error Signal. We calculate an X Error and a Y Error by simply subtracting the frame center from the box center.

Managing Jitter with a Control Deadband

If our object is perfectly centered, our Error is zero. If the object moves to the right, the X Error becomes a positive number. If it moves to the left, it becomes a negative number. The same logic applies vertically to our Y Error.

Now, you might think we should tell the servos to move every single time the Error is anything other than zero. But remember what we learned about camera sensors: pixels dance, light fluctuates, and your calculations will always have a tiny amount of natural mathematical noise. If you try to correct for every single fractional pixel change, your servos will constantly buzz, twitch, and jitter themselves to death.

To fix this, we implement an engineering safety margin called a Deadband. In this lesson, we establish a 40-pixel safety zone around the center of the frame.

  • If the object is within 40 pixels of the center, the error is too small to care about, and we tell the servos to sit perfectly still.

  • The moment the object drifts outside that 40-pixel window, our control logic triggers.

If the X Error is greater than 40, we decrement our pan angle by one degree to turn the camera toward the target, pass that new angle to our servo handler, and pause for a tiny fraction of a second (20 milliseconds) to give the mechanical gears time to physically move. If it’s negative, we increment the angle. We apply the exact same behavioral logic to our tilt servo using the Y Error.

Visualizing the System Matrix

To help us calibrate and troubleshoot this system, we overlay clear visual indicators directly onto our live video feed:

  • The Reticle: We draw a solid blue dot directly at our fixed frame center. This acts as our tracking crosshair.

  • The Target: We draw a large red circle directly over the center of our moving object’s bounding box.

When your system is working properly, you can physically watch the machine think. As you move an object around, the red circle moves away from the blue dot, the error threshold trips, the servos kick in, and the camera moves until the red circle swallows the blue dot once again.

Your Homework Assignment

You guys know the drill: watching me build a tracking rig doesn’t make you an automation engineer. You have to write the logic, feel the hardware move, and tune it yourself.

Here is your homework challenge for Lesson 28: Right now, our tracking logic uses what is called an incremental step controller. No matter how far away the object is from the center, the camera always moves at the exact same speed—one lazy degree at a time. If you move your target slowly, the camera keeps up. If you snap your target quickly across the room, the camera falls behind and loses it because it can’t accelerate.

Your assignment is to upgrade this control loop. Instead of stepping by a hardcoded value of 1, I want you to make the servo adjustment step proportional to the size of the error. If the object is close to the center, it should move gently by a fraction of a degree. If the object takes off like a rocket and creates a massive error signal, the camera should aggressively throw the servos open to catch up instantly.

Get your proportional tracking loops tuned, shoot a video showing your camera tracking a fast-moving object smoothly, upload it to YouTube, and share your link down in the comments below. See you guys in the next lesson!

AI On the Edge, Tutorial

AI on the Edge LESSON 23: Creating Regions of Interest (ROI) in OpenCV with Slicing

Welcome back, everyone! In this lesson, we are stepping into a foundational aspect of computer vision: manipulation of specific regions within a video frame.

Up to this point, we have been grabbing the full frame from our camera and performing operations on the entire image. But in real-world edge AI and robotics applications, processing every single pixel of a high-resolution frame is an absolute waste of compute power. If you want to detect a license plate, track a face, or monitor a specific sensor layout on a machine, you don’t need to look at the sky or the floor. You need to isolate a Region of Interest (ROI).

In this lesson, you will learn how to use Python’s powerful matrix slicing capabilities to chop up a frame, isolate specific quadrants, manipulate pixels inside an ROI, and display multiple synchronized windows across your desktop without crashing your system footprint.

The Core Concept: Image Slicing and ROIs

In OpenCV, an image frame isn’t just a visual picture—it is a standard NumPy array. A color frame is a 3D matrix structured by rows, columns, and color channels: [Rows, Columns, Channels] or [Height, Width, Color].

Because it is a standard array, we can use standard Python slicing notation to isolate any rectangular box we want:

ROI = frame[rowstart : rowend,  colstart : colend]

The .copy() Trap

When you slice a piece of an array in Python like ROI = frame[0:100, 0:100], Python does not create a new image in your RAM. It creates a view or a pointer back to the original frame. If you modify pixels inside that ROI, you will accidentally alter your original main camera frame!

To isolate a region and modify it independently without bleeding back into your primary frame, you must explicitly use the .copy() method:

Below is the complete code script we built during the video tutorial. Copy this code exactly into your Python environment, verify your geometry setups, and run it.

Homework Assignment

Alright, it is time to earn your stripes and see if you can fly with the big dogs. Your homework assignment is to take this foundation and build a dynamic tracking target box using the array geometry principles we just learned.

  1. Create a single main camera window (640 x 360).

  2. Draw an independent rectangular ROI box that starts directly in the dead center of the screen.

  3. Using your keyboard parameters (cv2.waitKey), program the system so that using the Arrow Keys (or ‘i’, ‘j’, ‘k’, ‘l’) smoothly updates variables to move the ROI box dynamically around the screen in real-time.

  4. Crucial Constraint: Do not let your boundary indices drift off the array! You must write conditional boundaries so that if your moving target hits the edge of your $640 \times 360$ boundary layout, it locks at the frame border and prevents an out-of-bounds index crash.

  5. In a separate output window, display only the contents of the moving target box in real-time grayscaled format.

Grab your morning coffee, fire up your code editor, write the script from scratch, and do not copy-paste code you don’t understand. Leave a link to your homework solution video in the YouTube comments section so I can see your progress!

AI On the Edge

AI on the Edge LESSON 22: Understanding Pictures and Video Frames as a Data Structure

Hey guys, Paul McWhorter here with TopTechBoy.com, and today we are diving into the heart of computer vision. We’ve been playing around with getting images from the camera, but have you ever stopped to actually look at what a picture is when it’s inside your computer’s memory?

If you want to be a master of AI on the Edge, you have to stop thinking about images as “pictures” and start seeing them as what they really are: a massive, organized grid of numbers.

What is a Picture, Really?

In this lesson, we are peeling back the curtain on how OpenCV and Python handle video frames. When we call piCam.capture_array(), we aren’t just taking a snapshot; we are pulling a data array into memory.

Think of it like a giant spreadsheet where every single cell is a pixel.

  • Dimensions: Your image has a width and a height, which correspond to the number of rows and columns in that array. It is important to remember the row designator comes first, then the column, [ R, C]

  • The Depth (The RGB Channels): It’s not just a flat 2D grid! Each “cell” in that grid is actually a little sub-array containing three values: Red, Green, and Blue. That is why we call it a 3D data structure.

Manipulating Data, Not Just Pixels

The magic happens when you realize you can reach into that array and change those numbers directly.

In the code we developed today, we aren’t just displaying video; we are performing data science on video frames. We explored how to:

  1. Access individual pixels: By referencing specific coordinates in our frame array, we can pull out the color data for a single spot.

  2. Draw shapes by modifying arrays: Notice how we don’t need a “draw square” function to put a box on the screen? We simply tell a slice of that array to equal [0, 0, 255]. We are literally changing the color values of those pixels to solid red.

  3. Regions of Interest (ROI): This is critical for AI. You don’t always need to look at the whole frame. We learned how to “slice” the array to isolate a Region of Interest. By carving out a smaller piece of that memory, we can perform operations—like converting to grayscale—on just that section, which saves a massive amount of processing power.

Why Does This Matter?

If you want to build a robot that recognizes objects or tracks faces, you need to understand this structure. AI models don’t “see” a cat; they see a mathematical representation of that cat’s pixel values. By learning how to slice, manipulate, and convert these arrays, you are learning the fundamental language of machine learning.

We are building the foundation here, folks. Once you get comfortable with how to manipulate these arrays, we are going to start doing some really cool stuff with image processing and filtering.

Dive into that code, change those array values, and see what happens when you mess with the dimensions! Don’t just run it—experiment with it.

I’ll see you guys in the next lesson!

In this lesson we developed the following code: