tttttttttttttttttttttttttttttttttttttttttttrssssss ttttttt ny nytt nr nrttttttt ny ntt nrttttttt zzzz bcorote to whol th ticunthss tidio tiostolonzzzz trougfht to ferr oy disectins it has dicomered brought to view by dissection it was discovered Tr a i n i n g iterations Figure 1: Text recognition from unaligned data. We present a method for recognising text in images without using any labelled data. This is achieved by learning to align the statistics of the predicted text strings, against the statistics of valid text strings sampled from a corpus. The figure above visualises the transcriptions as various characters are learnt through the training iterations. The model first learns the concept of {space}, and hence, learns to segment the string into words; followed by common words like {to, it}, and only later learns to correctly map the less frequent characters like {v, w}. The last transcription also corresponds to the ground-truth (punctuations are not modelled). The colour bar on the right indicates the accuracy (darker means higher accuracy).
ABSTRACTThis work presents a method for visual text recognition without using any paired supervisory data. We formulate the text recognition task as one of aligning the conditional distribution of strings predicted from given text images, with lexically valid strings sampled from target corpora. This enables fully automated, and unsupervised learning from just line-level text-images, and unpaired text-string samples, obviating the need for large aligned datasets. We present detailed analysis for various aspects of the proposed method, namely -(1) impact of the length of training sequences on convergence, (2) relation between character frequencies and the order in which they are learnt, (3) generalisation ability of our recognition network to inputs of arbitrary lengths, and (4) impact of varying the text corpus on recognition accuracy. Finally, we demonstrate excellent text recognition accuracy on both synthetically generated text images, and scanned images of real printed books, using no labelled training examples.