Ian J. Goodfellow, Yaroslav Bulatov, Julian Ibarz, Sacha Arnoud, Vinay Shet
2013 · arXiv
Recognizing arbitrary multi-character text in unconstrained natural photographs is a hard problem. In this paper, we address an equally hard sub-problem in this domain viz. recognizing arbitrary multi-digit numbers from Street View imagery. Traditional approaches to solve this problem typically separate out the localization, segmentation, and recognition steps. In this paper we propose a unified approach that integrates these three steps via the use of a deep convolutional neural network that operates directly on the image pixels. We employ the DistBelief implementation of deep neural networks in order to train large, distributed neural networks on high quality images. We find that the performance of this approach increases with the depth of the convolutional network, with the best performance occurring in the deepest architecture we trained, with eleven hidden layers. We evaluate this approach on the publicly available SVHN dataset and achieve over $96\%$ accuracy in recognizing complete street numbers. We show that on a per-digit recognition task, we improve upon the state-of-the-art, achieving $97.84\%$ accuracy. We also evaluate this approach on an even more challenging dataset generated from Street View imagery containing several tens of millions of street number annotations and achieve over $90\%$ accuracy. To further explore the applicability of the proposed system to broader text recognition tasks, we apply it to synthetic distorted text from reCAPTCHA. reCAPTCHA is one of the most secure reverse turing tests that uses distorted text to distinguish humans from bots. We report a $99.8\%$ accuracy on the hardest category of reCAPTCHA. Our evaluations on both tasks indicate that at specific operating thresholds, the performance of the proposed system is comparable to, and in some cases exceeds, that of human operators.
Metadata: arXiv. Abstracts retain their original authors’ rights.
Original HN comments, with labels from simple text rules. A label describes the whole comment and may miss context or irony. Read the discussion to make up your own mind. About the labels
7 selected excerpts · 7 total mentions
The state of the art has improved so much that for the older tasks, computers are arguably better than humans at them: https://arxiv.org/abs/1312.6082 So humans have to clear higher and higher hurdles to prove themselves human while the computers use the answers from the captchas to get closer to humans' capabilities.
Sure, there are a lot of potential things to do - either by enumerating towards the full "brute force" loss partway as I think you are hinting toward, or treating as a subsequences, or something else [0, 1, 2]. The point is, IMO RNNs do need some kind of structured loss (more than per step likelihood) to be competitive with HMM approaches using Viterbi decoding, in addition to some sequence smarts (such as beam search) in decoding. At least in NMT, enumerating the possible decisions as 3n + 1 is almost impossible since the softmax size is generally the memory bottleneck in training - and a bigger vocabulary is typically a huge win. It is more feasible in speech, but often…
your labels are themselves triphones and you end up with a pretty large vocabulary too. Figuring out how to get RNNs closer to on par with DNN-HMMs with Viterbi decoding (or full on sequence training [3]) through something like "deep fusion" with a language model (or something else) is something I am very interested in. [0] http://arxiv.org/abs/1312.6082 [1] http://arxiv.org/abs/1511.06456 [2] https://arxiv.org/abs/1511.04868 [3] http://www.danielpovey.com/files/2013_interspeech_dnn.pdf
We found that neural networks can solve CAPTCHAS much better than humans, 99.8% on the "hard" ReCAPTCHA instances: http://arxiv.org/pdf/1312.6082.pdf This is why visual recognition is just one of the signals you need to use to tell humans and computers apart http://googleonlinesecurity.blogspot.com/2014/04/street-view...
It's a new success for Deep Learning. It seems to be actually 99.8% accuracy according to their paper: http://arxiv.org/abs/1312.6082 That's one order of magnitude higher.
This is essentially the technology that was discussed here 3 months ago [1], and it links to the exact same article on arxiv, titled: "Multi-digit Number Recognition from Street View Imagery using Deep Convolutional Neural Networks". [2] They new addition to the article is that now they have tested the same type of NN on reCAPTCHA, and (perhaps unsurprisingly) it works. [1] - https://news.ycombinator.com/item?id=7015602 [2] - http://arxiv.org/abs/1312.6082v4.
I think the actual paper (linked at the end) is much more informative and interesting than this coverage: http://arxiv.org/abs/1312.6082 (I wouldn't call it blogspam, because it looks like they interviewed the researchers, but the summary leaves something to be desired)
Distorted text isn't even used in reCAPTCHA all that much anymore -- Google used a Deep Convolutional Neural Network and got over 90% accuracy in reading street numbers in Street View, and 99.8% accuracy in solving distorted text reCAPTCHA (http://arxiv.org/abs/1312.6082). Most users will just check a "I'm not a robot" box now; and if you do get a test, it will likely be a computer vision / image labeling problem: https://googleonlinesecurity.blogspot.com/2014/12/are-you-ro...
More from the same topics.
Sébastien Bubeck, Varun Chandrasekaran, Ronen Eldan +11
Artificial intelligence (AI) researchers have been developing and refining large language models (LLMs) that exhibit remarkable capabilities across a variety of domains and tasks, challenging our understanding of learning and cognition.
Tom B. Brown, Benjamin Mann, Nick Ryder +28
Recent work has demonstrated substantial gains on many NLP tasks and benchmarks by pre-training on a large corpus of text followed by fine-tuning on a specific task.
Chiyuan Zhang, Samy Bengio, Moritz Hardt +2
Despite their massive size, successful deep artificial neural networks can exhibit a remarkably small difference between training and test performance.
A. Pluchino. A. E. Biondo, A. Rapisarda
The largely dominant meritocratic paradigm of highly competitive Western cultures is rooted on the belief that success is due mainly, if not exclusively, to personal qualities such as talent, intelligence, skills, efforts or risk taking.