b7b1bd5c6f7892061a9915cd27d19360482d1fd8,lib/model/losses.py,,gmsd_loss,#Any#Any#,556

Before Change

    gms = tf.div(upper, lower)
    _mean, _var = tf.nn.moments(gms, axes=[1, 2], keep_dims=True)
    // single metric value per image in tensor [?, 1, 1]
    gmsd = tf.reduce_mean(tf.sqrt(_var), axis=-1)
    // need to expand to [?, height, width] dimensions for Keras ... modify to not be hard-coded
    return K.tile(gmsd, [1, 64, 64])


def ms_ssim(img1, img2, max_val=1.0, power_factors=(0.0517, 0.3295, 0.3462, 0.2726)):
    

After Change

    true_edge = scharr_edges(y_true, True)
    pred_edge = scharr_edges(y_pred, True)
    ephsilon = 0.0025
    upper = 2.0 * true_edge * pred_edge
    lower = K.square(true_edge) + K.square(pred_edge)
    gms = (upper + ephsilon) / (lower + ephsilon)
    gmsd = K.std(gms, axis=(1, 2, 3), keepdims=True)
    gmsd = K.squeeze(gmsd, axis=-1)
    return gmsd


def ms_ssim_calc(img1, img2, max_val=1.0, power_factors=(0.0517, 0.3295, 0.3462, 0.2726)):
    

Instances