tf2rl.tools package
Submodules
tf2rl.tools.img_tools module
- tf2rl.tools.img_tools.random_crop(input_imgs, output_size)
- Parameters
input_imgs – np.ndarray Images whose shape is (batch_size, width, height, channels)
output_size – Int Output width and height size.
Returns:
- tf2rl.tools.img_tools.center_crop(img, output_size)
- Parameters
img – np.ndarray Input image array. The shape is (width, height, channel)
output_size – int Width and height size for output image
Returns:
- tf2rl.tools.img_tools.preprocess_img(img, bits=5)
Preprocessing image, see https://arxiv.org/abs/1807.03039.
tf2rl.tools.vae module
- class tf2rl.tools.vae.VAE(*args, **kwargs)
Bases:
tensorflow.python.keras.engine.training.Model- __init__(latent_dim, inference_net, generative_net)
- sample(eps=None)
- encode(x)
- reparameterize(mean, logvar)
- decode(z, apply_sigmoid=False)
- compute_loss(x)
- compute_apply_gradients(x)
- tf2rl.tools.vae.log_normal_pdf(sample, mean, logvar, raxis=1)