Unrot logounrot
LoginGet Started for free
School and kids

AI Project for Class 10 CBSE: Ideas, Project Cycle, File Format and Viva

Satvik ParamkushamUpdated October 5, 202613 min read

Quick answer

A good AI project for Class 10 CBSE (subject code 417) solves one real problem linked to a Sustainable Development Goal, follows the AI Project Cycle from problem scoping to evaluation, and uses one of the three domains: statistical data, computer vision or NLP. In the 2026-27 curriculum, project work carries 10 marks plus 5 for its viva.

If you took Artificial Intelligence (subject code 417) as your skill subject in Class 10, your practical marks depend on three things: a practical file, a practical exam and a project (or a field visit or portfolio). The project is where most students lose easy marks, usually because they pick a topic that is too big or skip a stage of the AI Project Cycle.

This guide gives you the marks breakdown from the official CBSE curriculum, the six project cycle stages explained in plain words, 10 project ideas that fit the syllabus, a project file format your teacher will recognise, and the viva questions that come up most often.

One rule before you start: CBSE updates the curriculum every session. Always check the current year's 417 curriculum PDF on cbseacademic.nic.in, and follow your teacher's instructions where they differ from this page.

How many marks is the Class 10 AI project worth?

In the CBSE 2026-27 curriculum for Artificial Intelligence (code 417), the subject is out of 100: 50 for theory and 50 for practical and project work. The project itself is worth 10 marks, with another 5 for a viva on it.

Part C: practical and project workMarks
Practical file with minimum 15 programs15
Practical exam (statistical data, computer vision, NLP, advance Python)15
Viva voce5
Project work, field visit or student portfolio (any one)10
Viva voce on the project5
Total50

The theory side splits into 10 marks for employability skills (Part A) and 40 for the AI units (Part B). Notice that the project and its viva together make 15 marks, the same as the whole practical exam. A clear, honest project you can explain is worth real marks.

The curriculum also asks that projects relate to the Sustainable Development Goals (SDGs), so pick a problem you can connect to one of them, such as clean water, zero hunger or climate action.

What are the stages of the AI Project Cycle in Class 10?

The AI Project Cycle in the CBSE syllabus has six stages: problem scoping, data acquisition, data exploration, modelling, evaluation and deployment. Class 9 introduces them and Class 10 revisits them in Unit 1, so your project report should show every stage.

  1. Problem scoping: decide exactly what problem you are solving. CBSE uses the 4Ws problem canvas: Who has the problem, What the problem is, Where it happens, and Why solving it matters. End with a one-line problem statement.
  2. Data acquisition: collect reliable data. Name your source (a public dataset, a survey you ran, photos you took) and the features you need.
  3. Data exploration: clean the data and look at it with graphs (bar, line, scatter) to spot patterns and gaps.
  4. Modelling: choose a rule-based or learning-based approach and train a model, for example a decision tree or an image classifier.
  5. Evaluation: test the model on data it has not seen. Class 10 Unit 3 covers train-test split, accuracy, error, the confusion matrix, precision, recall and F1 score.
  6. Deployment: show how people would use it, even as a simple demo, poster or prototype.

Ethics runs through every stage. Ask whether your data is fair to everyone it affects, an idea the syllabus calls AI bias.

How do I choose an AI project topic for Class 10?

Choose a problem you can see in your own school, home or town, match it to one domain, and confirm you can get the data in a week. Small and finished beats big and half-done.

Ask yourself four questions:

  • Can I name the user? "Farmers in my district" is better than "everyone".
  • Can I get the data? A public dataset, your own photos or a class survey all count.
  • Which domain is it? Numbers and tables mean statistical data; photos mean computer vision; text or speech means NLP.
  • Can I show a result? An accuracy figure or a confusion matrix makes evaluation easy to present.

You do not have to code. The curriculum lists no-code tools such as Orange Data Mining and Teachable Machine, and teachers accept them. If you want to write Python, keep it to the libraries from your practicals: NumPy, pandas, Matplotlib and, if your teacher allows, scikit-learn.

10 AI project ideas for Class 10 CBSE

The first three are the sample projects named in the CBSE curriculum itself. The rest fit the same units and can each be linked to an SDG. Each one works with a no-code tool or with beginner Python.

  1. 1. Penguin species predictor (statistical data)

    Domain:
    Statistical data
    Tools:
    Orange or Python
    Time:
    1 week

    Predict whether a penguin is Adelie, Chinstrap or Gentoo from its bill length, flipper length and body mass. CBSE uses this exact Palmer Penguins case study in Unit 4, so your teacher will know it well. The data is free (it also loads in Python through seaborn). Build it in Orange Data Mining with a decision tree, or in Python with scikit-learn. What makes it stand out: explain which feature mattered most and show a confusion matrix. Upgrade: compare a decision tree with one other model and explain the difference in accuracy. SDG link: life on land and below water.

  2. 2. Coral bleaching early detector (computer vision)

    Domain:
    Computer vision
    Tools:
    Teachable Machine, Orange
    Time:
    1 week

    Classify reef photos as healthy or bleached. CBSE names this as its computer vision sample project and links the steps and dataset from the curriculum PDF. Train it in Teachable Machine or Orange with image embeddings, then test it on photos it has never seen. What makes it stand out: a clear explanation of why early detection matters for SDG 14 (life below water). Upgrade: add a third class such as "partly bleached" and show how accuracy changes when classes get harder to tell apart.

  3. 3. Product review sentiment analyser (NLP)

    Domain:
    NLP
    Tools:
    Orange or Python
    Time:
    1 week

    Sort reviews into positive and negative. Sentiment analysis is the NLP sample project in the CBSE curriculum, done in Orange with real review data, and the free IMDB movie review dataset on Hugging Face is another option. Show text normalisation (lowercasing, removing stopwords) and a bag of words, both Unit 6 topics. What makes it stand out: test it on reviews you write yourself, including sarcasm, and report where it fails. Upgrade: try reviews in Hinglish and explain why accuracy drops. Our sentiment analysis build shows the Python version.

    Step-by-step guide

  4. 4. Smart waste sorter

    Domain:
    Computer vision
    Tools:
    Teachable Machine
    Time:
    4 to 5 days

    Train a model to tell plastic, paper and food waste apart from webcam photos. The CBSE curriculum suggests a "Build a Smart Sorter" activity with Teachable Machine, so this extends work you may already have done in class. Take 50 to 100 photos per class at home, in different light. What makes it stand out: a deployment idea such as a labelled bin with a phone camera above it. Upgrade: add a "not sure" rule when confidence is below 70 percent. SDG link: responsible consumption (SDG 12).

  5. 5. Air quality explorer for Indian cities

    Domain:
    Statistical data
    Tools:
    Python, pandas, Matplotlib
    Time:
    1 week

    Explore daily AQI readings for Indian cities using the Air Quality Data in India dataset on Kaggle, then predict whether a day will be "poor" or worse from PM2.5 and PM10 readings. Plot monthly averages to show which months are worst. What makes it stand out: data exploration with clear graphs, which is exactly what the practical exam tests. Upgrade: compare your own city with one other and suggest one action for your school on high-AQI days. SDG link: good health (SDG 3).

  6. 6. Plant leaf disease identifier

    Domain:
    Computer vision
    Tools:
    Teachable Machine
    Time:
    1 week

    Help farmers spot disease early by classifying leaf photos as healthy or diseased. The PlantVillage dataset on Kaggle has thousands of labelled leaf images; pick one crop, such as tomato, and two or three classes. Train in Teachable Machine, which handles images without any code. What makes it stand out: the farmer use case and a clear note on what the model cannot do (it has only seen lab-style photos). Upgrade: test it on real leaves from a garden and report the accuracy gap. Our image classification project shows the Python route.

    Step-by-step guide

  7. 7. SMS spam detector

    Domain:
    NLP
    Tools:
    Python, scikit-learn
    Time:
    1 week

    Classify text messages as spam or not spam using the SMS Spam Collection dataset from the UCI Machine Learning Repository. Turn each message into a bag of words or TF-IDF scores, both in the Unit 6 syllabus, and train a simple classifier. What makes it stand out: a real problem every family in India faces, and a perfect example for explaining precision versus recall (missing a spam message is less harmful than blocking a bank OTP). Upgrade: collect 20 spam messages from your family's phones (with permission, personal details removed) and test on them.

  8. 8. Handwritten digit recogniser

    Domain:
    Computer vision
    Tools:
    scikit-learn or Teachable Machine
    Time:
    4 to 5 days

    Recognise handwritten digits 0 to 9, linked to the learning outcome on how computers identify alphabets and handwriting. Use the small digits dataset that ships with scikit-learn (8 by 8 pixel images), or train Teachable Machine on photos of your classmates' handwriting. What makes it stand out: you can explain pixels, grayscale values and resolution, all computer vision theory topics. Upgrade: show the confusion matrix and explain which digits get mixed up most, for example 3 and 8.

  9. 9. School helpdesk chatbot: script bot vs smart bot

    Domain:
    NLP
    Tools:
    Python
    Time:
    1 week

    Build a chatbot that answers questions about your school timetable, fees dates or library rules. Make two versions: a script bot that matches keywords, and a smarter version that compares the question with stored questions using a bag of words. The script bot versus smart bot comparison is a Class 10 NLP topic, so this project doubles as viva preparation. Upgrade: log 20 real questions from classmates and report how many each version answered correctly. The Python chatbot guide walks through the code.

    Step-by-step guide

  10. 10. Study hours and marks predictor

    Domain:
    Statistical data
    Tools:
    Google Sheets, Orange or Python
    Time:
    1 to 2 weeks

    Run an anonymous survey in your class (study hours, sleep hours, screen time, last test score) and train a decision tree to predict whether a student scores above 75 percent. Decision trees appear in the learning outcomes, and the survey gives you your own data acquisition story. What makes it stand out: a strong ethics section on consent, keeping names out of the data, and why a model should never be used to label a student. Upgrade: show what happens to accuracy when you remove one feature at a time.

What should a Class 10 AI project file contain?

A Class 10 AI project file should follow the AI Project Cycle, one section per stage, with your evidence (canvas, graphs, screenshots, results) in each. Use this order unless your teacher gives a template:

  1. Cover page: project title, your name, class, roll number, school, session.
  2. Certificate signed by the teacher, then acknowledgement.
  3. Index with page numbers.
  4. Introduction and the SDG your project supports.
  5. Problem scoping: the completed 4Ws canvas and a one-line problem statement.
  6. Data acquisition: source, features, number of rows or images.
  7. Data exploration: two or three graphs, each with one sentence on what it shows.
  8. Modelling: tool used, approach, screenshots of training.
  9. Evaluation: accuracy, confusion matrix and what went wrong.
  10. Deployment: how a real user would use it.
  11. Ethics, limitations and future scope.
  12. Bibliography with every dataset and tool named.

Keep the practical file separate. It needs at least 15 programs, and the curriculum suggests programs such as adding two lists, mean, median and mode with NumPy, line and scatter charts, reading a CSV and displaying an image with Python.

What viva questions are asked on the Class 10 AI project?

Viva questions test whether you understood your own project, so expect "why" questions about each stage. Prepare a one-line answer to each of these:

  • What problem does your project solve, and for whom? Which SDG does it support?
  • Fill in your 4Ws canvas for me out loud.
  • Where did your data come from, and how much did you use?
  • Is your model rule-based or learning-based? Is it supervised or unsupervised?
  • What is the difference between AI, ML and deep learning? (Read AI vs ML vs deep learning once before the viva.)
  • Why did you split the data into training and testing sets?
  • What is your accuracy, and why is accuracy alone not enough? Explain precision and recall with your project.
  • What is a confusion matrix? Read yours aloud.
  • What would happen if your training data came from only one city or one kind of photo?
  • How would you improve the project with more time?

What mistakes lose marks in a Class 10 AI project?

The most common mistakes are copying a project from the internet, skipping evaluation, and choosing a problem with no data. Watch for these:

  • Downloading a finished project. Examiners ask about your data and your results, and a copied file shows within two questions.
  • No evaluation. "The model works" earns little. Test on unseen data and write down the accuracy.
  • Too broad a problem. "AI for pollution" is a theme. "Predict poor-AQI days in Delhi from PM2.5" is a project.
  • Testing on training data. The accuracy looks perfect and means nothing, a mistake called overfitting when the model memorises instead of learning.
  • Missing ethics. One paragraph on bias, privacy and consent shows the thinking the curriculum asks for.
  • No sources. Name every dataset and tool in the bibliography.

Want more ideas at a slightly easier level? See AI projects for school students.

Frequently asked questions

Which is the best AI project for Class 10 CBSE?

The best project is one you can finish and explain. The three CBSE sample projects (penguin species prediction, coral bleaching detection and sentiment analysis) are safe choices because teachers know them. For extra credit, pick a local problem tied to an SDG, such as AQI in your city or a waste sorter for your school.

Can I do the Class 10 AI project without coding?

Yes. The CBSE 417 curriculum itself lists no-code tools such as Orange Data Mining and Teachable Machine for statistical data, computer vision and NLP work. You still need to show every stage of the AI Project Cycle, including evaluation with accuracy, in your report.

What are the 6 stages of the AI Project Cycle?

The six stages in the CBSE syllabus are problem scoping, data acquisition, data exploration, modelling, evaluation and deployment. Problem scoping uses the 4Ws canvas (who, what, where, why), and evaluation uses measures such as accuracy, precision, recall and F1 score.

How many programs are needed in the Class 10 AI practical file?

The 2026-27 CBSE curriculum asks for a practical file with a minimum of 15 programs, worth 15 marks. Suggested programs include list operations, mean, median and mode with NumPy, line and scatter charts, reading CSV files and reading images with Python. Check your school's list too.

Is a field visit or portfolio allowed instead of a project?

Yes. In the 2026-27 curriculum, Part C lets students do any one of project work, a field visit or a student portfolio for 10 marks, followed by a 5-mark viva. A portfolio needs a minimum of 5 activities. Your school decides which options it offers, so confirm with your teacher.

Keep learning

AI ProjectsAI Projects for School Students: 11 Ideas for Class 6 to 12AI ProjectsAI Projects for Kids: 12 Fun, Safe Ideas for Ages 8 to 14AI ProjectsAI Projects for Beginners: 11 Easy Python Ideas, Easiest FirstAI ProjectsSentiment Analysis Project in Python: From TF-IDF Baseline to TransformersLearn the conceptWhat Is Artificial Intelligence? A Simple Explanation for Beginners and StudentsLearn the conceptAI vs Machine Learning vs Deep Learning: How They Differ (With Examples)Learn the conceptComputer VisionLearn the concept/best/ai-courses-for-kids

References

All ai projects