Write-it-Yourself: Empowering Blind People to Independently Fill-out Paper Forms
Syed Masum Billah, Shirin Feiz, Vikas Ashok, Roy Shilkrot, I. V. Ramakrishnan · 2019 · Proceedings of the 16th International Web for All Conference (W4A) · doi:10.1145/3315002.3332447
Summary
This demonstration paper introduces WiYG (Write-it-Yourself Guide), a system that enables blind people to independently fill out printed paper forms such as bank checks, consent forms, and restaurant receipts — a task that has until now been impossible without sighted assistance. The core problem is that blind people cannot identify where form fields are located on a printed page or determine what information each field requires. Existing assistive technologies have focused exclusively on reading printed materials (through OCR-based tools like KNFB Reader, Seeing AI, and wearable devices like FingerReader and OrCam), with no support for the complementary task of writing. WiYG consists of three components: a standard signature guide (a credit-card-sized rectangular card with a cutout opening that blind people already use to write in straight lines) augmented with ArUco fiducial markers for computer vision tracking; a pocket-sized custom 3D-printed smartphone attachment that props the phone at an angle to capture a full sheet of paper while keeping the touchscreen accessible; and a smartphone application that uses computer vision to continuously detect the signature guide's position and generate real-time audio navigation instructions directing the user to each form field.
Key findings
A user study with 13 blind participants demonstrated that WiYG enabled users to correctly fill out form fields at the right locations with an accuracy of up to 89.5%. The system was tested with four different form types: a consent form, a letter-sized bank check, a restaurant receipt, and a standard check. Human annotators evaluated whether each field was filled in the correct location, with correctly placed entries highlighted in green and misplaced ones in red. The workflow is straightforward: the user attaches the 3D-printed apparatus to their phone, places the phone in front of the paper, aligns the signature guide to the top-left corner for calibration, then taps the screen to receive voice instructions for moving the signature guide to each form field sequentially. After writing in each field, the user taps to receive guidance to the next field until the form is complete. The system builds on well-established computer vision algorithms for marker detection and requires only a standard smartphone — no specialized hardware beyond the inexpensive 3D-printed attachment.
Relevance
This work addresses a fundamental gap in assistive technology: while significant progress has been made in enabling blind people to read printed materials, the equally important task of writing on printed documents has been largely neglected. Despite the dominance of digital transactions, paper forms remain ubiquitous in daily life — bank checks, medical forms, tax returns, contracts, and receipts all require handwritten entries and signatures. The inability to fill these out independently forces blind people into dependence on sighted assistance for routine tasks, with implications for privacy (sharing financial or medical information with helpers) and autonomy. WiYG's approach is practical and low-cost: it leverages a standard smartphone, a 3D-printed attachment that can be produced cheaply, and a familiar tool (the signature guide) that many blind people already own. The pocket-sized design means the system is portable for use in banks, offices, and restaurants. For accessibility practitioners and developers, this paper demonstrates how combining existing technologies (computer vision, 3D printing, audio guidance) in novel ways can address real-world accessibility barriers that have persisted despite decades of assistive technology research focused primarily on reading.
Tags: blindness · assistive technology · computer vision · 3D printing · writing assistance · paper forms · independence · OCR · audio guidance