Skip to content

An unofficial implementation of W-Net for crowd counting.

License

Notifications You must be signed in to change notification settings

ZhengPeng7/W-Net-Keras

Repository files navigation

W-Net-Keras

Unofficial implementation of W-Net for crowd counting in Keras.


Paper:

Results now:

On dataset ShanghaiTech B

Under development...

MAE MSE Mean of Frobenius Norm MAPE PSNR SSIM
7.85 12.17 6.75e-7 6.46% 29.27 0.93

Dataset:

Training Parameters:

  1. Loss = MSE * 1000 + BCE * 10;

  2. Optimizer = Adam(lr=1e-4, decay=5e-3);

  3. Batch size: 1;

  4. Data augmentation: Flip horizontally online randomly;

  5. Patch: No patch;

  6. Batch normalization: No BN layers at present;

  7. Weights: Got best weights in epoch248(250 epochs in total), and here is the loss records:

    Loss_records

  8. Prediction example:

    example

Run:

  1. Download dataset;
  2. Data generation: run thegenerate_datasets.ipynb .
  3. Run the main.ipynb to train, test, analyze and evaluate the image quality.

About

An unofficial implementation of W-Net for crowd counting.

Topics

Resources

License

Stars

Watchers

Forks

Packages

No packages published