FACLSTM: ConvLSTM with focused attention for scene text recognition

Keyphrases

• Convolutional Long Short-term Memory (ConvLSTM) 100%
• Focus of Attention 100%
• Scene Text Detection 100%
• Text-to-image 75%
• Prediction Problems 50%
• Fully Connected 50%
• Structural Information 25%
• 2D Image 25%
• Benchmark Dataset 25%
• Feature Vector 25%
• Convolution Operation 25%
• Spatial Data 25%
• Image Input 25%
• Attention Mechanism 25%
• Large Margin 25%
• Feature Map 25%
• Spatial Correlation 25%
• Severe Damage 25%
• Feature Area 25%
• Scene Text 25%
• Noisy Text 25%
• Curved Text 25%
• Recognizer 25%
• Sequential Prediction 25%
• Spatio-temporal Prediction 25%
• Additional Characters 25%
• Sequential Features 25%
• Factorized Convolution 25%
• Sequence Prediction 25%

Engineering

• Prediction Problem 100%
• Experimental Result 50%
• Input Image 50%
• Limitations 50%
• Feature Vector 50%
• Critical Role 50%
• Spatial Correlation 50%
• Severe Damage 50%
• Recognizer 50%

Computer Science

• Long Short-Term Memory Network 100%
• Focused Attention 100%
• Experimental Result 14%
• Attention (Machine Learning) 14%
• Feature Vector 14%
• Spatial Correlation 14%
• Recognizer 14%
• Focus Attention 14%