Functions — github.com/12dmodel/deep_motion_mag

CodeHub MCPcopy Create free account

hub / github.com/12dmodel/deep_motion_mag / functions

Functions53 in github.com/12dmodel/deep_motion_mag

⨍ Functions 53 ◇ Types & classes 2

↓ 14 callersMethodrun

Magnify a video in the two-frames mode. Args: checkpoint_dir: checkpoint directory. vid_dir: directory containing vid

↓ 11 callersFunctionconv2d

(input_, output_dim, ks=4, s=2, stddev=0.02, padding='SAME', name="conv2d", activation_fn=None,

↓ 6 callersFunctionimread

(path, is_grayscale=True)

↓ 5 callersMethodload

(self, checkpoint_dir, loader=None)

↓ 5 callersFunctionresidual_block

(x, output_dim, ks=3, s=1, name='residual_block')

↓ 4 callersFunction_bytes_feature

convert_3frames_data_to_tfrecords.py:16

↓ 4 callersMethod_encoder

↓ 4 callersFunctionload_train_data

(image_path, gray_scale=True, is_testing=False)

↓ 3 callersFunctionL1_loss

↓ 3 callersFunctionpreprocess_image

(image, train_config)

preprocessor.py:5

↓ 2 callersMethod_decoder

(self, texture_enc, shape_enc)

↓ 2 callersFunctionget_possion_noise

Add poisson noise. This function approximate posson noise upto 2nd order. Assume images were in 0-255, and converted to the range of -1 to 1.

preprocessor.py:16

↓ 2 callersMethodimage_transformer

(self, image_a, image_b, amplifi

↓ 2 callersFunctioninverse_transform

↓ 2 callersFunctionmkdir

↓ 1 callersMethod_build_IIR_filtering_graphs

↓ 1 callersMethod_build_feed_model

↓ 1 callersMethod_build_training_graph

(self, train_config)

↓ 1 callersFunction_float_feature

convert_3frames_data_to_tfrecords.py:13

↓ 1 callersFunctioncenter_crop

(x, crop_h, crop_w, resize_h=64, resize_w=64)

↓ 1 callersFunctionconvert_dataset

(data_dir, out_name, color=False)

convert_3frames_data_to_tfrecords.py:20

↓ 1 callersFunctionexpand_dims_1_to_4

Expand dimension from 1 to 4. Useful for multiplying amplification factor.

↓ 1 callersFunctionimsave

(images, size, path)

↓ 1 callersMethodinference

Run Magnification on two frames. Args: frameA: path to first frame frameB: path to second frame amplifica

↓ 1 callersFunctionmain

convert_3frames_data_to_tfrecords.py:74

↓ 1 callersFunctionmain

↓ 1 callersFunctionmerge

↓ 1 callersFunctionpreproc_gaussian_noise

preprocessor.py:38

↓ 1 callersFunctionpreproc_poisson_noise

preprocessor.py:33

↓ 1 callersFunctionread_and_decode_3frames

(filename_queue, im_size=(512, 512, 1))

data_loader.py:33

↓ 1 callersFunctionres_decoder

(activation, layer_dims, out_channels, num_resblk)

↓ 1 callersFunctionres_encoder

(image, layer_dims, num_resblk)

↓ 1 callersMethodrun_temporal

Magnify video with a temporal filter. Args: checkpoint_dir: checkpoint directory. vid_dir: directory containing video

↓ 1 callersMethodsave

(self, checkpoint_dir, step)

↓ 1 callersFunctionsave_images

(images, size, image_path)

↓ 1 callersMethodsetup_for_inference

Setup model for inference. Build computation graph, initialize variables, and load checkpoint.

↓ 1 callersMethodtrain

(self, train_config)

↓ 1 callersFunctiontransform

(image, npx=64, is_crop=True, resize_w=64)

(self, sess, name, arch_config)

(self, maxsize=50)

Method_loss_function

(self, a, b, train_config)

Functionbatch_norm

(x, name="batch_norm")

Functiondeconv2d

(input_, output_dim, ks=4, s=2, stddev=0.02, name="deconv2d", activation_fn=None, weights_regular

Functionget_gaussian_noise

preprocessor.py:29

Functionget_image

(image_path, image_size, is_crop=True, resize_w=64, is

Functioninstance_norm

(input, name="instance_norm")

(input_, output_size, scope=None, stddev=0.02, bias_start=0.0, with_w=False)

(x, leak=0.2, name="lrelu")

Functionmerge_images

Functionpreproc_color

preprocessor.py:45

Functionread_and_decode

(filename_queue, im_size=(512, 512, 1))

data_loader.py:4

Functionres_manipulator

(enc_a, enc_b, amplification_factor, layer_dims,