Street view numberdetection is falling into the category of the natural scene text recognitionproblem which is quite different from printed character or handwrittenrecognition problem. Research in this field was started in 90’s but still, itis considered as an unsolved problem. As I mentioned earlier that itsdifficulties are due to fonts variation, scales, rotations, low lights etc.
In earlier years to dealwith this matter sequentially, character classification by sliding window *from 4 or connected components * from 4 mainly used. After that wordprediction can be done by predicting character classifier in left to rightmanner. Recently * from 4 segmentation method guided by supervised classifieruse where words can be recognized through a sequential beam search. * from 4But none of this can help to solve the street view recognition problem.In recent worksconvolutional neural networks proves its capabilities more accurately to solveobject recognition task. * from 4 Some research has done with CNN to tacklescene text recognition tasks. * from 4 on that studies CNN shows its hugecapability to represent all types of character variation in the natural sceneand till now it is holding this highly variability. Analysis with convolutionalneural network stars at early 80’s and it successfully applied for handwrittendigit recognition in 90’s * from 4 After that with the increasing availabilityof computer resources, training sets, advance algorithm * from 3 and dropouttraining *from 3 must lead to many successes using deep convolutional neuralnetworks.
Previously CNN usedmainly for detecting a single object from an input image. It was quitedifficult to isolate each character from a single image and identify them.Goodfellow * from 4 solve this problem by a deep large CNN directly to modelthe whole image and a simple graphical model as top inference layer. The rest of the paper isdesigned in section III Convolutional neural network architecture, section IVExperiment, Result, and Discussion and Future Work and Conclusion in section V