This is a network for text recognition scenario. It consists of VGG16-like backbone and bidirectional LSTM encoder-decoder. The network is able to recognize case-insensitive alphanumeric text (36 unique symbols).
|Accuracy on the alphanumeric subset of ICDAR13||0.8818|
|Text location requirements||Tight aligned crop|
Shape: [1x1x32x120] - An input image in the format [BxCxHxW], where:
Note that the source image should be tight aligned crop with detected text converted to grayscale.
The net outputs a blob with the shape [30, 1, 37] in the format [WxBxL], where:
The network output can be decoded by CTC Greedy Decoder or CTC Beam Search decoder.
[*] Other names and brands may be claimed as the property of others.