6b48055fdfe3c628833ed26ca394ef2fb1d97f82,niftynet/layer/resampler.py,ResamplerLayer,_resample_nearest,#ResamplerLayer#Any#Any#,58

Before Change

        spatial_coords = self.boundary_func(
            tf.round(sample_coords), in_spatial_size)
        batch_ids = tf.reshape(
            tf.range(batch_size), [batch_size] + [1] * (out_spatial_rank + 1))
        batch_ids = tf.tile(batch_ids, [1] + out_spatial_size + [1])
        output = tf.gather_nd(
            inputs, tf.concat([batch_ids, spatial_coords], -1))

        if self.boundary == "ZERO":
            scale = 1. / (tf.constant(in_spatial_size, dtype=tf.float32) - 1)

After Change

        spatial_coords = self.boundary_func(
            tf.round(sample_coords), in_spatial_size)
        output = tf.stack([
            tf.gather_nd(img, coords) for (img, coords) in
            zip(tf.unstack(inputs), tf.unstack(spatial_coords))])

        if self.boundary == "ZERO":
            scale = 1. / (tf.constant(in_spatial_size, dtype=tf.float32) - 1)
            mask = tf.logical_and(

Instances