Skip to content

timespacce/ocr_challenge

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

19 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

OCR Challenge with Convolutional Neural Network

Data preparation

  1. 34318 train and 5509 test images with different resolutions.
  2. Resize to 64x128, normalize each channel to [0, 1] and keep HWC format.

Model

  1. 3 x Convolution Blocks (CONV - CONV - CONV - BATCH NORM - LEAKY RELU - DROPOUT)
  2. CONV(label_length)
  3. Dense(hidden_Size)
  4. Dense(vocab_size)
  5. Softmax(-1)

Training

  1. Adaptive Momentum Estimation with 1e-3 and Constant Scheduling Strategy
  2. batch_size = 512 on DGX A100 for 20 epochs (250 seconds)
  3. Reshuffle at each epoch
  4. Minimize Categorical Cross Entropy between one-hot-encoded network output and one-hot-encoded label

Evaluation

  1. Accuracy defined as number of correctly predicted pictures.

    Model Parameters : [MB] Train Test
    32-512 4,039,403 : 15.4MB 93.2 % 70.1 %

About

Convolutional Neural Network for character recognition

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages