1M+
Routes Planned
Love Exploring
Offline-first travel app with AI itineraries and hidden gem discovery
- React Native
- Maps
- Travel
- Offline
AI & Emerging
TextBehind
Computer vision SaaS that places text behind subjects in 2 seconds
AIComputer VisionSaaSWeb
Project Snapshot
- Client
- TextBehind Inc
- Industry
- Creative & Design Tools
- Platform
- Web SaaS
- Timeline
- 5 months
- Our Role
- AI model integration + SaaS frontend
The Challenge
The 'text-behind-subject' visual effect — where typography appears to weave between the subject and the background — had become a signature look in premium design work. But achieving it required 15-20 minutes of manual masking per image in Photoshop, putting it out of reach for most creators. Existing AI tools that attempted automation were either too slow (30+ seconds per image) or too inaccurate (visible artefacts around hair and complex edges). TextBehind needed something fast enough to feel instant and accurate enough to be production-ready.
Our Solution
We built a computer vision pipeline combining Meta's Segment Anything Model for initial segmentation with a custom depth estimation model for layering decisions. The pipeline runs on GPU-accelerated AWS Lambda functions, processing images in parallel. Our key innovation was a refinement pass specifically trained on hair, fine clothing edges, and transparent materials — the cases where generic segmentation fails. The frontend is a React SaaS with real-time typography controls, so designers can adjust font, size, and position with immediate visual feedback on the processed result.
Key Features
What We Built
2-Second AI Processing
Custom GPU pipeline processes subject segmentation and text layering in under 2 seconds for most images.
Real-Time Typography Controls
Adjust font, size, weight, tracking, and position with live preview — no re-processing required per adjustment.
Batch Processing (100+ images)
Queue up to 100 images for parallel batch processing — ideal for marketing campaigns and content libraries.
Brand Kit Integration
Upload brand fonts, colours, and presets — apply your brand identity to any image in one click.
API Access for Developers
RESTful API with webhook callbacks for integrating TextBehind processing into any design or CMS workflow.
Template Library (500+)
500+ pre-designed text layout templates sorted by use case — social, advertising, editorial, and packaging.
Tech Stack
- Python
- FastAPI
- React
- TensorFlow
- OPOpenCV
- SESegment Anything Model
- λAWS Lambda
- CLCloudFront
Screenshots
App in Action
Scroll to see all →
“What took our design team 20 minutes per image now takes 2 seconds. The accuracy around hair and fine edges is remarkable — it's better than what we were doing manually.”
Daniel Park
Head of Design, TextBehind Inc
Why 200+ Companies Choose iMOBDEV Over the Rest
AI-Native Approach
We don't retrofit AI; we design for it from day one — architecture, data, and UX.
On-Time, Every Time
Agile sprints with clear milestones, transparent reporting, and zero deadline surprises.
Transparent Pricing
No hidden costs. Fixed-price or dedicated team models — you choose what fits.
Global Team, Local Focus
India-based delivery excellence with US and UAE client engagement offices.
Enterprise-Grade Security
ISO 27001-compliant, NDA-ready, GDPR-aware — secure by design at every layer.
Ongoing Partnership
We're a long-term partner, not a one-and-done vendor. We grow as you grow.
Ready to Build Something Like This?
Tell us your challenge. Free consultation, no commitment.