Skip to content
Adept Sense

One identity journey.
Every check connected.

Meet SenseAI, the identity intelligence suite by Adept Sense. Bring NID OCR, liveness and Face Match together in one episode—with every result in one review portal.

Request a demo Explore the episode

Built for the documents, faces and everyday realities of Bangladesh.

Fictional adult woman illustrating a SenseAI identity journey Nusrat Rahman Fictional identity
Illustrative sequence
Sample National ID card for Nusrat Rahman, DEMO-0001, marked SAMPLE / NOT VALID
NID OCRRead the document
LivenessCheck presence
Face MatchCompare identity
One episodeEvery signal, together
SenseAI product suiteNID OCRLiveness SDKFace MatchLanguage & name APIs

Three checks.
One complete picture.

An identity validation episode connects the document, the person’s presence and the reference comparison. Select a stage to follow the evidence.

One identity validation episode

Start with what the card says.

Read Bangla and English NID fields, distinguish father and spouse names, and transliterate Bangla-only names. Front and back evidence remain connected to the same episode.

Explore NID OCR

Every episode.
One review workspace.

Move from the operational overview to a single journey. Inspect OCR evidence, liveness and video, and the government reference comparison side by side.

SenseAI review portalWatch the portal walkthrough
Edited actual portal showing fictional Nusrat Rahman’s NID OCR, liveness evidence, and Face Match together. The hero sample NID and its younger grayscale portrait appear as the document and Government Identity Database reference, with matching OCR fields.
Actual portal interface · fictional identity and illustrative results. Preview only.
See the operational overview Task counts, review states and correlation queue
Actual portal overview with OCR, liveness, Face Match, correlated episodes and work queue summaries
Supplied portal screenshot. Counts reflect the captured interface, not performance benchmarks.

Evidence stays connected. Reviewers keep the context.

See the portal in a demo

Built for the hard cases.

Real identity checks involve worn documents, changing appearances and imperfect capture conditions. SenseAI brings purpose-built processing to each stage.

NID OCR that understands
more than characters.

Read Bangla and English NID cards—even when the challenge starts with a photocopy, a hard crease or faded ink. Intelligent recognition and recovery help extract useful fields from difficult scans.

  • Names, in context. Identify whether the card lists a father or spouse. Transliterate Bangla-only names, including parents’ names, into English.
  • Corrections you can teach. Train recognition for recurring patterns and known card errors. In some countries, identity cards include “Late” before a deceased mother’s or father’s name, while the Government Identity Database omits it. This difference can cause verification to fail. Configure normalization to remove “Late” before matching.
  • Both sides, connected. Read front and back fields, including PDF417 barcodes and TD1 machine-readable zones used across card formats.
  • Reuse the work. Cached OCR results can be returned when the same card is submitted again.
Explore OCR integration
Structured identity Fictional sample
নুসরাত রহমানNUSRAT RAHMANDEMO-0001 · 12 Jun 1996
Father / পিতা
রহমান আলীRahman Ali
Mother / মাতা
নাসরিন আক্তারNasrin Akter
Relationship identified
Father
Evidence
Front + back
Configurable normalization example

Late রহমান আলী Rahman Ali

Sample fields correspond to the fictional portal episode above.

Younger adult version of a fictional woman, with uncovered hair, in a low-detail degraded reference portrait
Degraded reference illustration350x230 pixel
Later portrait of the same fictional woman wearing a hijab, with natural age-related facial changes
Current capture illustrationYears later · wearing a headscarf
Younger clean-shaven adult version of a fictional man in a low-detail degraded reference portrait
Degraded reference illustration350x230 pixel
Later portrait of the same fictional man with a full beard and natural age-related facial changes
Current capture illustrationYears later · with a beard

Fictional people. Illustrations of age, appearance and input quality.
Not measured match or recovery results.

People change.
Identity has a structure.

A government reference may be years older than today’s selfie. Age, a head covering or a newly grown beard can make comparison harder. Low resolution and compression add another layer of difficulty.

  • Recover before comparing. SenseAI’s image recovery system processes degraded reference images before the matching stage.
  • Look beyond surface differences. Its AI-driven architecture evaluates facial structure alongside overall appearance to determine likeness.
  • Keep the evidence in view. Inspect the reference, submitted selfie and matching result within the same episode, with review policies suited to your business.
Explore Face Match

On-device awareness.
Server-side judgment.

Liveness starts in the user’s environment and finishes with a server decision. The SDK and detection engine work together through a workflow you control.

  1. Guide and inspect on the device.

    Prompt the user to perform configured tasks. Check lighting, environment and image quality, and look for banding, moiré and other clues associated with video playback and spoof attempts.

  2. Send the capture for a final verdict.

    Record a five-second video and send it with the on-device findings for server evaluation. The identity check compares the selfie and video within this capture flow.

  3. Configure, review and improve.

    Add or remove tasks remotely, control the workflow and tune the decision matrix to business requirements. Administrators can tag mistakes and use those examples to train for new conditions and customer demographics.

Explore the Liveness SDK
Liveness SDKIllustrative sequence
Fictional woman illustrating SDK capture
Look straight ahead

Example task. Your server determines the tasks and their order.

No camera access. This is a product illustration.

Names travel.
Keep their meaning connected.

Use the language and name APIs on their own, or alongside your identity workflow.

নুসরাতNusrat

Bangla–English transliteration

Reconcile names across scripts. Transliterate individual names or batches through the API.

Explore transliteration
Nusrat Name-based estimate

Gender estimation

Return a name-based estimate and confidence information. This describes patterns in names and does not establish a person’s gender identity.

Explore name analysis

Fits your stack.
Follows your flow.

Go from a sample request to an integration you understand. Explore inputs, inspect responses, and connect the services from your backend.

Mobile SDKsiOS & Android Web integrationHosted capture Server APIsJSON & multipart
Open the documentation
name-transliterationRequest example
Bangla ↔ EnglishTry sample data

Bring the whole
identity journey together.

Explore SenseAI for your documents, your users and your review policies.

Talk to Adept Sense Explore the API playground

Portal preview

Fictional identity episode portal preview

One episode. The complete picture.

Captured portal interface · fictional identities · illustrative results

Scroll the image to inspect details
Step 1 of 5

See the whole operation

OCR tasks, liveness attempts and Face Match results come together in one operational overview.

Start with the big picture.