🤖 It's Machine Learning time!

Let's get a glimpse of the world of computer vision this week — through a practical application with a problem that hits close to home.

Manually cropping faces from photos using photo editing software is time-consuming. The task this week is, to automate this process using machine learning-based face detection to produce square, face-centered images for an ID-making software.

⚙️ Task breakdown:
- Detect faces in each photo using OpenCV
- Crop each face into a square image, ensuring the face is centered with a padding
- Output square images suitable for ID-making software

This streamlines ID photo preparation, saving time and ensuring consistent output.

🎒 Resources:
- OpenCV for Processing
- Get started with this code

✏️ Recommended Learning:
- loadImage (), image (), save ()
- mask (), copy (), PGraphics

⚡️ Bring the Thunder!
- support for slightly rotated photos
- support for mass photo processing

🪁 Submit your code in the comments!

@WeeklyCoder | Week 78

🩺 Symptom-Based Disease Predictor

Let's use oral exam data to make predictions.

Given this CSV file that maps diseases to symptoms on a severity scale, build a program that asks users to input their symptom severities and returns the most likely disease from the table.

Each row is a disease, each column is a symptom (rated 0–3). Scale-Meaning.
0 - Absent
1 - Mild
2 - Moderate
3 - Severe

🥅 Goal: Build a program that interacts with users to guess the most likely disease from their reported symptoms.

🦶 Steps:
- load and parse the CSV file
- prompt the user to enter their severity for each symptom (0–3)
- compare the user's input against all diseases
- output the most likely disease

✏️ Recommended Learning:
- Table, loadTable ()
- getRowCount (), getColumnCount ()
- getString (), getInt ()
- sortKeys (), sortValues ()
- .keyArray ()
- similarity/distance calculation (sum of absolute differences)

🪁 Submit your code in the comments!

@WeeklyCoder | Week 79

👉 Read this first: Tough-Love

📸 Manual Photo Crop Tool

Last time, we stepped into the world of computer vision — building a tool to automatically detect and crop faces from photos for ID-making purposes. It works great ... except when ML didn’t.

Not every photo played nice with automated face detection. Sometimes it missed the face or cropped poorly.

So this week, we’re flipping the script — let's build the user custom and manual control to crop faces from photos.

🥅 Goal: Build a Manual Crop Tool where a user can manually position the photo behind a fixed square frame to get the perfect crop.

🦶 Steps: here

⚙️ Key Features:
- Panning: on the image (click-drag or arrow keys)
- Zooming: in/out using mouse wheel or +/-

⚡️ Bring the Thunder: here

✏️ Recommended Learning:
- loadImage(), image(), translate(), scale(), rotate()
- mask (), copy(), save()
- mouseDragged(), mouseWheel(), keyPressed()
- PGraphics

🪁 Submit your code in the comments!

@WeeklyCoder | Week 80

📊 Fuel Theft Detection

This week, let's analyze vehicle telemetry to tackle a critical problem: detecting fuel theft. By processing GPS and fuel data, we can identify suspicious fuel level changes.

🥅 Goal: Build a Fuel Theft Detector that processes a CSV of vehicle telemetry data to identify stops and flag fuel changes as theft or fillings.

We'll use this real-world data as input.

Real life is messy. Sensors can give readings that are all over the place when a vehicle is moving uphill or down hill or when it changes.

So to do this, we'll only consider fuel level changes as either thefts/fillings if they happen while the vehicle is stationary (speed = 0 kph). We'll take as a stop a duration of ≥ 3 minutes.

Filling if Fuel Diff > 1.0 and stop duration ≥ 3 minutes and Theft if Fuel Diff < -1.0

⚡️ Bring the Thunder:
- Output a summary report (events and amounts) to a csv file.
- Include the associated GPS coordinates to indicate where theft or fillings may have happened.

✏️ Recommended Learning:
- loadTable(), saveTable(), nf
- SimpleDateFormat
- Loops and conditionals
- .replace (), .contains ()

🪁 Submit your code in the comments!

@WeeklyCoder | Week 81

Ever found yourself copying something useful, only to accidentally overwrite it with something else seconds later?

I've been there. Windows has a clipboard history feature built in — BUT it's limited, buggy, and doesn't quite do enough for real-world use.

👎 The built-in Windows clipboard:

- Stores only 25 items
- Clears everything on restart
- No search or categorization
- No support for files
- Misses to record things sometimes
- Does not have searchability
- Does not have categorization
- and a lot more

But we can do better — and we will.

🥅 Goal: To build a better clipboard manager. BetterClip will have:

- Unlimited history
- Searchability
- Persistent storage across reboots
- Auto-deletion of old records
- Restore on startup
- Support for text, images, files, URLs
- Tabs for categorization [All | Texts | Images | Files | URLs | +]
- Option to pin
- Global hotkey to launch
- Incognito mode to temporarily disable recording

✏️ Recommended Learning:
- java.awt.Toolkit | pyperclip
- JIntellitype for Global Hotkeys
- the object Object and DataFlavor
- saveStrings ()
- List, LinkedHashMap

🪁 Submit your code in the comments!

@WeeklyCoder | Week 82

DJ Robot

Music in factories boosts morale, productivity, and focus. But manual playlist management often leads to forgotten starts, missed pauses, and repetitive songs.

🥅 Goal: This week, let's build a customizable, offline music scheduler for factory Bluetooth PA systems — with the following features:

- automated start/stop at clock-in/out
- pause during lunch breaks (or custom lunch playlist)
- day-specific playlists
- easy way to add/remove songs in playlists

⚡️ Bring the Thunder:
- schedule overrides no playback on Sundays or specific dates (religious holidays or during no work days)
- bluetooth PA system integration such that it looks for the PA system before playing music
- song rotation to avoid repetition

✏️ Recommended Learning:
- Minim library for audio playback
- java.util.Timer for scheduling automation
- java.io.File for playlist/file management
- List, HashMap, +
- JIntellitype for special media keys (Play, Pause, Stop, +)

🪁 Submit your code in the comments!

@WeeklyCoder | Week 83

🧾 StitchPay: Statement Consolidator

You'd think banks such as CBE, established as they are, would include something basic as a Payer name in a bank statement they send. But you'd think wrong. 🤦‍♂️

So, this week, let's extend our Receipt Detective challenge of Week 59 to build a Statement Consolidator that takes in an incomplete bank statement and gives back something complete.

🥅 Goal: Given the account number of a Receiver, build a program that takes in a list of Reference numbers and gives back a complete statement as .csv file.

The transaction section of a bank statement should sensibly include these — but as a start, let's build a payer name fetcher.

🗒 Format here.

🦶 Build on top of this example.

⚡️ Bring the Thunder!
- So generate a complete Bank Statement as a CSV from a list of Reference numbers.

✏️ Recommended Learning:
- URL, InputStream, File, FileOutputstream
- PDFbox, PDFTextStripper
- .openStream (), .deleteOnExit ()

🪁 Send your submissions in the comments!

@WeeklyCoder | Week 84

Inclusive Modular Knapsack

A school is looking to buy every type of item for their students — notebooks, pencils, and erasers — but it has capped the total spend at 2,000 Br (t = total).

🥅 Goal: This week, build a solver that finds how many packs of each supply type can be bought so that:
- every item type is included at least once
- quantities are unlimited
- total cost is no more than 't' Br but can be a little under

Example:
- Notebook pack → 150 Br
- Pencil pack → 90 Br
- Eraser pack → 60 Br
- Target budget t = 2000 Br

⚡️ Bring the Thunder: here

✏️ Recommended Learning:
- Coin change / knapsack dynamic programming
- Residue arithmetic to handle multiples
- Data structures: arrays, maps, +

@WeeklyCoder | Week 85