No abstract

Text localization from the digital image is the first step for the optical character recognition task. Conventional image processing based text localization performs adequately for specific examples. Yet, a general text localization are only archived by recent deep-learning based modalities. Here we present document Text Localization Generative Adversarial Nets (TLGAN) which are deep neural networks to perform the text localization from digital image. TLGAN is an versatile and easy-train text localization model requiring a small amount of data. Training only ten labeled receipt images from Robust Reading Challenge on Scanned Receipts OCR and Information Extraction (SROIE), TLGAN achieved 99.83% precision and 99.64% recall for SROIE test data. Our TLGAN is a practical text localization solution requiring minimal effort for data labeling and model training and producing a state-of-art performance.

Our previous studies have shown that introducing Si doping in quantum dots (QDs) can help QD solar cells achieve higher voltage. However, this improvement came at the cost of current loss. In this work, we continue to investigate the cause of the current loss and propose a method to recover it without compromising the voltage. Photoluminescence measurements have confirmed that optimizing the thickness of the GaAs layers in the i-region can lead to strong current gain (~14%) with minimal voltage loss (<3%) and alteration of the QD quality. The capacitance–voltage measurement results support that the current gain mainly originates from the increased depletion width.