FingerReader2.0

Abstract

People with Visual Impairments (PVI) experience greater difficulties with daily tasks, such as supermarket shopping. Identifying and purchasing an item proves challenging for PVI. Using a user-centered design process, we understand the difficulties PVI encounter in their daily routines. Consequently, the previous FingerReader model was elevated to a new level. In contrast, FingerReader2.0 incorporates a highly integrated hardware design, as it is standalone, wearable, and not tethered to a computer. Software-wise, the prototype utilizes a deep learning system, relying on a hybrid, an on-board and a cloud-based model. The advanced design significantly extends the range of mobile assistive technology, particularly for shopping purposes. This paper presents the findings from interviews, several iterative studies, and a field study in supermarkets to demonstrate the FingerReader2.0's enhanced capabilities for those with varied levels of visual impairment.

FetchAid: Making Parcel Lockers More Accessible to Blind and Low Vision People With Deep-learning Enhanced Touchscreen Guidance, Error-Recovery Mechanism, and AR-based Search Support

“…Global parcel volume has reached 159 billion, with projections to hit 216 to 300 billion by 2027 2 . "Last-Mile Delivery, " the final step in parcel delivery from a hub to its destination, has been a bottleneck in express delivery efficiency 3 .…”

“…Even if they successfully navigate the touchscreen, they must still locate the open compartment and safely approach it without tripping or colliding with its door [32]. as computer vision and crowdsourcing for camera aiming and touchscreen information [18], wearable devices for object recognition and localization [3,7], and systems like NavCog for turn-by-turn navigation assistance [1,22,40,47], there is a need for an innovative, systematic solution that holistically addresses all three challenges, without requiring extra hardware and online help.…”