NewEDWartens Lab is open: 18 free machines (opens in a new tab)

Online · Free

Computer Vision Internship

8 short weeks. Free. Start any date.

Start free Plan my pace

By Wartens Ltd (UK), delivered through EDWartens.

A 4 by 4 grey image drawn as a grid of numbers from 10 to 250, with shape (4, 4).
  • Free
  • Online
  • Any start date
  • 1 to 8 weeks

Who it suits

Made for beginners.

You do not need to have written image code. The course starts with what a picture is to a computer and builds up to detection and inspection in small steps.
  • Students of computer science, electronics, electrical, mechatronics and mechanical engineering.
  • Anyone curious how a camera can count items, read a label or spot a defect.
  • Read the notes on a phone, a tablet or a laptop.
A colour pixel with three numbers is turned into one gray number using weights for red, green and blue.

What you will learn

Eight weeks, one picture each

Every week is open from day one. Each lesson takes a few minutes.
  1. A 4 by 4 grey image drawn as a grid of numbers from 10 to 250, with shape (4, 4).

    Week 1

    How computers see: pixels, colour, images

    Explain how a picture is stored as numbers, read a pixel by row and column, and say why OpenCV colour order is BGR.

  2. A colour pixel with three numbers is turned into one gray number using weights for red, green and blue.

    Week 2

    Image processing with OpenCV

    Convert an image to grayscale, threshold it to black and white, blur away noise and clean up the result with erode and dilate.

  3. A row of pixel values that jumps from dark to light, with the gradient peaking at the jump.

    Week 3

    Features, edges and contours

    Turn a picture into edges, turn the edges into outlines, and count and measure the shapes in it with OpenCV.

  4. Two paths: hand-written rules turn an image into a decision, while labelled examples train a model that turns an image into a class and a confidence.

    Week 4

    Classification with deep learning (CNNs)

    Explain how a convolutional network sorts images into classes, read its training curves, and train a small classifier of your own.

  5. A photo area with two boxes, each labelled with a class and a confidence score, and a table listing the same two detections.

    Week 5

    Object detection and counting

    Read a detector's output, work out the overlap of two boxes as IoU, explain confidence and thresholds, and count objects without counting any twice.

  6. A conveyor with a part, a trigger sensor, a camera with light above it, a computer and a PLC connected to a rejector.

    Week 6

    Machine vision for quality inspection

    Pick a light and a camera set-up for a given inspection, work out a lens and a pixel size, and tell which defects need rules and which need a trained model.

  7. A bar chart of 1000 images in which 950 are good and only a few show a scratch, a dent or a crack.

    Week 7

    Data, labelling and accuracy

    Collect and label a small image set with clear rules, split it fairly, and read the results of a pass or fail model without being fooled by a high score.

  8. A camera sends images to an edge box, which decides pass or fail and signals the PLC to reject the part.

    Week 8

    Deploying vision at the edge

    Plan a vision inspection from camera to reject, and say how it is timed, handed to the PLC and watched once it is running.

A 200 millisecond time budget split into capture, preprocessing, model, decision and PLC signal, with time to spare.

Practice, if you like

Try it on a few images.

Each week ends with a worked exercise on a few pictures, with the answer or a good result given. Skip them if you prefer.
  • Free tools: Google Colab, OpenCV and Teachable Machine in your browser.
  • Your own photos of objects, or any image you have the right to use.
  • A mini project in week 8: a small vision model ready for a small device.
  • Best on a laptop or tablet. A phone works for reading.
A row of pixel values that jumps from dark to light, with the gradient peaking at the jump.
A photo area with two boxes, each labelled with a class and a confidence score, and a table listing the same two detections.

Your pace

Choose a start date and a length.

We show your weekly plan. You can change it later.

Any length from 1 to 8 weeks. Start on any date from today. Retry the weekly checks and the final quiz as often as you like.

Plan my internship
Two paths: hand-written rules turn an image into a decision, while labelled examples train a model that turns an image into a class and a confidence.

Certificate and documents

Want proof? Add it later.

Free learners keep full access. A verifiable certificate of completion is an optional paid document, issued by Wartens Ltd (UK). Anyone can check it with the QR code or number.
A bar chart of 1000 images in which 950 are good and only a few show a scratch, a dent or a crack.
How verification works

Questions

Quick answers

Is the Computer Vision Internship really free?

Yes. Joining is free and so are all eight weeks, the weekly checks and the final quiz. You pay only if you choose an optional add-on.

How long can I take?

This internship runs for 1 to 8 weeks. You pick a start date and can change your plan later. All eight weeks are open from day one, so a short plan moves fast and a long one gives you room.

What do I need?

A laptop or tablet, a free Google account for Colab, and a free AI assistant. A phone camera is handy for taking your own test pictures. Every tool used is free or has a free tier, so there is nothing to buy. The notes also read well on a phone.

What will I actually do?

You read short notes with diagrams, watch a few videos and take an easy check each week. Each week ends with a worked exercise, such as turning a picture grey, finding edges, counting objects or judging how accurate a classifier is. Week 8 is about running a small model close to the camera.

What do I get at the end?

A certificate of completion is an optional paid document, issued by Wartens Ltd (UK). It has a QR code and a number, so anyone can verify it at edwartens.com/verification. The lessons stay free, and you can finish without buying anything.

Is this a paid position?

No. It is training, not employment. Nothing is paid to you, and nothing here promises work with any company.

Who wrote the course?

OFORO, the Wartens AI company, wrote the notes and exercises. OFORO builds the AI behind Wartens' engineering and training work. The internship is run by Wartens Ltd (UK) and delivered through EDWartens.

What about privacy and wrong answers?

Vision models make mistakes in poor light and on things they have not seen, and the course teaches you to measure that. Photos of people are personal data, so use pictures of objects or images you have the right to use. Never upload other people's photos or confidential images to a public tool.

Your first lesson is free.

Start free
  • No card to start
  • Start any date
  • Verify with a QR code
Start free