FaSTExt: Fast and Small Text Extractor
This work addresses the need for efficient text extraction in real-world applications by offering a more compact alternative to large deep convolutional networks, though it is incremental in its improvements.
The paper tackles the problem of text detection in natural images by proposing a small and fast method that reduces model complexity using depthwise separable convolutions and other techniques, achieving parameter counts from 1.58 to 10.59 million while maintaining effectiveness on public datasets.
Text detection in natural images is a challenging but necessary task for many applications. Existing approaches utilize large deep convolutional neural networks making it difficult to use them in real-world tasks. We propose a small yet relatively precise text extraction method. The basic component of it is a convolutional neural network which works in a fully-convolutional manner and produces results at multiple scales. Each scale output predicts whether a pixel is a part of some word, its geometry, and its relation to neighbors at the same scale and between scales. The key factor of reducing the complexity of the model was the utilization of depthwise separable convolution, linear bottlenecks, and inverted residuals. Experiments on public datasets show that the proposed network can effectively detect text while keeping the number of parameters in the range of 1.58 to 10.59 million in different configurations.