ML Engineer for Document Scan Quality Model
Budget: ₹1,500 – ₹12,500 INR
Job Title: Machine Learning Engineer for Document Scanner Quality Model (Flutter Integration)
Project Overview:
We seek a skilled Machine Learning Engineer to develop a lightweight ML model that can detect and assess the quality of document or ID card scans in a live camera feed, integrated with a Flutter application. The model will replace traditional image processing libraries (e.g., OpenCV) and provide real-time user guidance such as “Center document,” “Hold still,” and “Avoid reflections.” The primary goal is to output accurate document detection, corner localization, and image quality metrics without OCR.
Key Responsibilities:
Design and build a compact ML model (e.g., CNN or transformer-based) for:
Detecting the presence and location of a rectangular document in camera frames.
Assessing image quality attributes: sharpness (blur), glare/reflections, framing/centering.
Train and validate the model on a diverse dataset representing real-world document scan conditions.
Convert the trained model to a mobile-friendly format (e.g., TensorFlow Lite) optimized for on-device inference in Flutter apps.
Create an API or interface for seamless Flutter integration, providing outputs used to trigger user guidance overlays and auto-capture decisions.
Collaborate with Flutter developers to ensure smooth integration while meeting performance and latency requirements.
Provide documentation on model usage, limitations, and suggestions for future improvements.
Project Overview:
We seek a skilled Machine Learning Engineer to develop a lightweight ML model that can detect and assess the quality of document or ID card scans in a live camera feed, integrated with a Flutter application. The model will replace traditional image processing libraries (e.g., OpenCV) and provide real-time user guidance such as “Center document,” “Hold still,” and “Avoid reflections.” The primary goal is to output accurate document detection, corner localization, and image quality metrics without OCR.
Key Responsibilities:
Design and build a compact ML model (e.g., CNN or transformer-based) for:
Detecting the presence and location of a rectangular document in camera frames.
Assessing image quality attributes: sharpness (blur), glare/reflections, framing/centering.
Train and validate the model on a diverse dataset representing real-world document scan conditions.
Convert the trained model to a mobile-friendly format (e.g., TensorFlow Lite) optimized for on-device inference in Flutter apps.
Create an API or interface for seamless Flutter integration, providing outputs used to trigger user guidance overlays and auto-capture decisions.
Collaborate with Flutter developers to ensure smooth integration while meeting performance and latency requirements.
Provide documentation on model usage, limitations, and suggestions for future improvements.