BUILD THE DATA.BUILD THEINTELLIGENCE.
A National AI Dataset Creation & Computer Vision Hackathon
Organized by the Department of Computer Science and Engineering, Ramco Institute of Technology
in association with RIT IEEE CS Student Chapter and RIT GFG Student Chapter
DataGenesis 2026 is a two-round national-level hackathon that challenges participants to create high-quality, India-centric datasets from scratch and transform them into intelligent computer vision models. Unlike conventional hackathons that provide ready-made datasets, DataGenesis places data creation, curation, and AI development at the heart of the challenge.
OPEN TO EVERYONE.
Open to students, researchers, professionals, and AI enthusiasts.
Participants from any academic stream or discipline are welcome.
No age restriction.
Maximum 3 members per team.
Participants from across India are encouraged to participate.
ROUND 01 — DATASET DEVELOPMENT
Teams will select one topic from the official dataset theme list and create an original, machine-learning-ready dataset.
TEAMS WILL:
- 01Select a dataset theme from the list given below.
- 02Prepare a dataset development plan.
- 03Capture original images under diverse conditions.
- 04Annotate and organize the images.
- 05Prepare complete dataset documentation.
- 06Publish the dataset on Kaggle.
- 07Submit the Kaggle link and required documentation through the official submission form.
DATASET QUALITY REVIEW
Submissions will be evaluated for:
- ▸Dataset planning
- ▸Image quality
- ▸Image diversity
- ▸Annotation accuracy
- ▸Dataset organization
- ▸Documentation
- ▸Overall machine-learning readiness
The review committee may request corrections or improvements before final qualification.
Only teams that meet the required quality standards will advance to Round 2.
IN-PERSON AT RIT.
Qualified teams will bring their Round 1 dataset to RIT and develop an object detection and recognition model from scratch.
TEAMS WILL:
- ▸Preprocess their dataset
- ▸Design their own model architecture
- ▸Implement the model
- ▸Train the model from scratch
- ▸Evaluate model performance
- ▸Demonstrate inference
- ▸Present their technical solution
THE FROM-SCRATCH CHALLENGE
To ensure originality, participants cannot use:
- ✕External or benchmark datasets
- ✕Pre-trained models or weights
- ✕Transfer learning
- ✕Foundation models
- ✕Existing trained object detection models
Teams must develop and train their model using their own Round 1 dataset with randomly initialized weights.
CHOOSE YOUR DOMAIN.
Participants can select one theme from the official list.
Indian Culture & Heritage
- Confectionery of India
- Traditional Indian Musical Instruments
- Indian Handloom Textures & Patterns
- Indian Leaf Plates
- Indian Toys and Games
- Traditional Indian Footwear
- Traditional Indian Lamps (Diyas & Oil Lamps)
- Indian Temple Bells and Ritual Objects
- Indian Terracotta and Clay Artifacts
- Traditional Indian Baskets and Woven Crafts
- Indian Pottery and Earthenware
- Traditional Indian Jewellery Designs
- Traditional Indian Self-Defence Implements
- Indian Attire Through the Ages
- Traditional Indian Games & Entertainment
Indian Agriculture & Natural Resources
- Exclusive Fruits & Vegetables from Indian Markets
- Indian Millets, Pulses, and Grains
- Indian Seeds and Seed Varieties
- Indian Medicinal Leaves and Herbs
Specialized Computer Vision Dataset
- Wrist Veins
HOW YOU'RE JUDGED.
ROUND 1 — DATASET DEVELOPMENT
ROUND 2 — AI MODEL DEVELOPMENT
₹30,000*
CASH PRIZE POOL
* Prize pool based on an estimated participation of 60 participants. Final prize allocation may vary based on actual participation.
REGISTER VIA GOOGLE FORMS
Submit your team and dataset through the official Google Form after curating your dataset. Round 1 participation is subject to the official registration and submission guidelines.
Applies to teams qualifying for Round 2.
- ✓Breakfast
- ✓Lunch
- ✓Supper
- ✓Snacks and refreshments
- ✓Participation certificate
- ✓Hackathon swag
Accommodation will be available on prior request and will be subject to additional charges and availability.