-
Notifications
You must be signed in to change notification settings - Fork 121
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Default data has only 1 channel. Fix the reshape issue for this defau…
…lt setting.
- Loading branch information
rensijie
committed
Dec 18, 2015
1 parent
c89e514
commit c706b2a
Showing
3 changed files
with
344 additions
and
344 deletions.
There are no files selected for viewing
228 changes: 114 additions & 114 deletions
228
applications/Shepard_CNN/Shepard_super_res/shepard_sr_x2_train.m
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -1,114 +1,114 @@ | ||
| %-------------------------------------------------------------------------------------------------------- | ||
| % The system is created based on the principles described in the following paper | ||
| % Jimmy SJ. Ren, Li Xu, Qiong Yan, Wenxiu Sun, "Shepard Convolutional Neural Networks", | ||
| % Advances in Neural Information Processing Systems (NIPS 2015) | ||
| % email: [email protected] | ||
| %-------------------------------------------------------------------------------------------------------- | ||
| addpath applications/Shepard_CNN/Shepard_super_res/ | ||
| addpath utils/ | ||
| addpath cuda/ | ||
| addpath mem/ | ||
| addpath layers/ | ||
| addpath layers_adapters/ | ||
| addpath optimization/ | ||
| addpath pipeline/ | ||
| addpath data/ | ||
|
|
||
| clearvars -global config; | ||
| clearvars -global mem; | ||
| clear gen_mask_patch_cat_idx_for_super_res; | ||
| clear; | ||
| global config mem; | ||
| shepard_sr_x2_configure(); | ||
| init(0); | ||
|
|
||
| load('data/Shepard_CNN/Shepard_super_res/x2/val_1ch/val_1'); | ||
|
|
||
| %images_t = reshape(images_t, size(images_t,1), size(images_t,2), 1, size(images_t,3)); | ||
| %labels_t = reshape(labels_t, size(labels_t,1), size(labels_t,2), 1, size(labels_t,3)); | ||
|
|
||
| perm = randperm(size(images_t, 4)); | ||
| images_t = images_t(:,:,:,perm); | ||
| labels_t = labels_t(:,:,:,perm); | ||
| test_samples = config.NEW_MEM(images_t(:,:,:,1:1000)); | ||
| test_labels = config.NEW_MEM(labels_t(:,:,:,1:1000)); | ||
|
|
||
| mask = config.NEW_MEM([1 0; 0 0]); | ||
| mask = repmat(mask, config.input_size(1)/2, config.input_size(2)/2, config.chs); | ||
| mask = repmat(mask, 1,1,1,config.batch_size); | ||
|
|
||
| count = 0; | ||
| cost_avg = 0; | ||
| epoc = 0; | ||
| points_seen = 0; | ||
| display_points = 5000; | ||
| save_points = 50000; | ||
| max_grad = 1; | ||
| fprintf('%s\n', datestr(now, 'dd-mm-yyyy HH:MM:SS FFF')); | ||
| for pass = 1:10 | ||
| for p = 1:50 | ||
| load(strcat('data/Shepard_CNN/Shepard_super_res/x2/train_1ch/patches_', num2str(p), '.mat')); | ||
|
|
||
| %images = reshape(images, size(images,1), size(images,2), 1, size(images,3)); | ||
| %labels = reshape(labels, size(labels,1), size(labels,2), 1, size(labels,3)); | ||
|
|
||
| perm = randperm(20000); | ||
| images = images(:,:,:,perm); | ||
| labels = labels(:,:,:,perm); | ||
| train_imgs = config.NEW_MEM(images); | ||
| train_labels = config.NEW_MEM(labels); | ||
|
|
||
| for i = 1:size(train_labels, 4) / config.batch_size | ||
| points_seen = points_seen + config.batch_size; | ||
| in = train_imgs(:,:,:,(i-1)*config.batch_size+1:i*config.batch_size); | ||
| %in = train_labels(:,:,:,(i-1)*config.batch_size+1:i*config.batch_size); | ||
| out = train_labels(:,:,:,(i-1)*config.batch_size+1:i*config.batch_size); | ||
| out = out((size(in, 1) - config.output_size(1)) / 2 + 1:(size(in, 1) - config.output_size(1)) / 2 + config.output_size(1), ... | ||
| (size(in, 2) - config.output_size(2)) / 2 + 1:(size(in, 2) - config.output_size(2)) / 2 + config.output_size(2), :, :); | ||
|
|
||
|
|
||
| % make the mask list | ||
| mask_li = {}; | ||
| mask_li{1} = mask; | ||
| % operate the training pipeline | ||
| op_train_pipe_with_mask(in.*mask, mask_li, out); | ||
| % update the weights | ||
| config.UPDATE_WEIGHTS(); | ||
|
|
||
| if(cost_avg == 0) | ||
| cost_avg = config.cost; | ||
| else | ||
| cost_avg = (cost_avg + config.cost) / 2; | ||
| end | ||
|
|
||
| % display point | ||
| if(mod(points_seen, display_points) == 0) | ||
| count = count + 1; | ||
| fprintf('%d ', count); | ||
| end | ||
| % save point | ||
| if(mod(points_seen, save_points) == 0) | ||
| fprintf('\n%s', datestr(now, 'dd-mm-yyyy HH:MM:SS FFF')); | ||
| epoc = epoc + 1; | ||
| test_cost = 0; | ||
| for t = 1:size(test_samples, 4) / config.batch_size | ||
| t_label = test_labels(:,:,:,(t-1)*config.batch_size+1:t*config.batch_size); | ||
| t_label = config.NEW_MEM(t_label((size(in, 1) - config.output_size(1)) / 2 + 1:(size(in, 1) - config.output_size(1)) / 2 + config.output_size(1), ... | ||
| (size(in, 2) - config.output_size(2)) / 2 + 1:(size(in, 2) - config.output_size(2)) / 2 + config.output_size(2), :)); | ||
|
|
||
| op_test_pipe_with_mask(test_samples(:,:,:,(t-1)*config.batch_size+1:t*config.batch_size).*mask, mask_li, t_label); | ||
|
|
||
| test_out = gather(mem.output); | ||
| test_cost = test_cost + config.cost; | ||
| end | ||
| test_cost = test_cost / (size(test_samples, 4) / config.batch_size); | ||
| fprintf('\nepoc %d, training avg cost: %f, test avg cost: %f\n', epoc, cost_avg, test_cost); | ||
|
|
||
| save_weights(strcat('applications/Shepard_CNN/Shepard_super_res/results/shepard_layer_x2/epoc', num2str(epoc), '.mat')); | ||
|
|
||
| cost_avg = 0; | ||
| end | ||
| end | ||
| end | ||
| end | ||
|
|
||
| %-------------------------------------------------------------------------------------------------------- | ||
| % The system is created based on the principles described in the following paper | ||
| % Jimmy SJ. Ren, Li Xu, Qiong Yan, Wenxiu Sun, "Shepard Convolutional Neural Networks", | ||
| % Advances in Neural Information Processing Systems (NIPS 2015) | ||
| % email: [email protected] | ||
| %-------------------------------------------------------------------------------------------------------- | ||
| addpath applications/Shepard_CNN/Shepard_super_res/ | ||
| addpath utils/ | ||
| addpath cuda/ | ||
| addpath mem/ | ||
| addpath layers/ | ||
| addpath layers_adapters/ | ||
| addpath optimization/ | ||
| addpath pipeline/ | ||
| addpath data/ | ||
|
|
||
| clearvars -global config; | ||
| clearvars -global mem; | ||
| clear gen_mask_patch_cat_idx_for_super_res; | ||
| clear; | ||
| global config mem; | ||
| shepard_sr_x2_configure(); | ||
| init(0); | ||
|
|
||
| load('data/Shepard_CNN/Shepard_super_res/x2/val_1ch/val_1'); | ||
|
|
||
| images_t = reshape(images_t, size(images_t,1), size(images_t,2), 1, size(images_t,3)); | ||
| labels_t = reshape(labels_t, size(labels_t,1), size(labels_t,2), 1, size(labels_t,3)); | ||
|
|
||
| perm = randperm(size(images_t, 4)); | ||
| images_t = images_t(:,:,:,perm); | ||
| labels_t = labels_t(:,:,:,perm); | ||
| test_samples = config.NEW_MEM(images_t(:,:,:,1:1000)); | ||
| test_labels = config.NEW_MEM(labels_t(:,:,:,1:1000)); | ||
|
|
||
| mask = config.NEW_MEM([1 0; 0 0]); | ||
| mask = repmat(mask, config.input_size(1)/2, config.input_size(2)/2, config.chs); | ||
| mask = repmat(mask, 1,1,1,config.batch_size); | ||
|
|
||
| count = 0; | ||
| cost_avg = 0; | ||
| epoc = 0; | ||
| points_seen = 0; | ||
| display_points = 5000; | ||
| save_points = 50000; | ||
| max_grad = 1; | ||
| fprintf('%s\n', datestr(now, 'dd-mm-yyyy HH:MM:SS FFF')); | ||
| for pass = 1:10 | ||
| for p = 1:50 | ||
| load(strcat('data/Shepard_CNN/Shepard_super_res/x2/train_1ch/patches_', num2str(p), '.mat')); | ||
|
|
||
| images = reshape(images, size(images,1), size(images,2), 1, size(images,3)); | ||
| labels = reshape(labels, size(labels,1), size(labels,2), 1, size(labels,3)); | ||
|
|
||
| perm = randperm(20000); | ||
| images = images(:,:,:,perm); | ||
| labels = labels(:,:,:,perm); | ||
| train_imgs = config.NEW_MEM(images); | ||
| train_labels = config.NEW_MEM(labels); | ||
|
|
||
| for i = 1:size(train_labels, 4) / config.batch_size | ||
| points_seen = points_seen + config.batch_size; | ||
| in = train_imgs(:,:,:,(i-1)*config.batch_size+1:i*config.batch_size); | ||
| %in = train_labels(:,:,:,(i-1)*config.batch_size+1:i*config.batch_size); | ||
| out = train_labels(:,:,:,(i-1)*config.batch_size+1:i*config.batch_size); | ||
| out = out((size(in, 1) - config.output_size(1)) / 2 + 1:(size(in, 1) - config.output_size(1)) / 2 + config.output_size(1), ... | ||
| (size(in, 2) - config.output_size(2)) / 2 + 1:(size(in, 2) - config.output_size(2)) / 2 + config.output_size(2), :, :); | ||
|
|
||
|
|
||
| % make the mask list | ||
| mask_li = {}; | ||
| mask_li{1} = mask; | ||
| % operate the training pipeline | ||
| op_train_pipe_with_mask(in.*mask, mask_li, out); | ||
| % update the weights | ||
| config.UPDATE_WEIGHTS(); | ||
|
|
||
| if(cost_avg == 0) | ||
| cost_avg = config.cost; | ||
| else | ||
| cost_avg = (cost_avg + config.cost) / 2; | ||
| end | ||
|
|
||
| % display point | ||
| if(mod(points_seen, display_points) == 0) | ||
| count = count + 1; | ||
| fprintf('%d ', count); | ||
| end | ||
| % save point | ||
| if(mod(points_seen, save_points) == 0) | ||
| fprintf('\n%s', datestr(now, 'dd-mm-yyyy HH:MM:SS FFF')); | ||
| epoc = epoc + 1; | ||
| test_cost = 0; | ||
| for t = 1:size(test_samples, 4) / config.batch_size | ||
| t_label = test_labels(:,:,:,(t-1)*config.batch_size+1:t*config.batch_size); | ||
| t_label = config.NEW_MEM(t_label((size(in, 1) - config.output_size(1)) / 2 + 1:(size(in, 1) - config.output_size(1)) / 2 + config.output_size(1), ... | ||
| (size(in, 2) - config.output_size(2)) / 2 + 1:(size(in, 2) - config.output_size(2)) / 2 + config.output_size(2), :)); | ||
|
|
||
| op_test_pipe_with_mask(test_samples(:,:,:,(t-1)*config.batch_size+1:t*config.batch_size).*mask, mask_li, t_label); | ||
|
|
||
| test_out = gather(mem.output); | ||
| test_cost = test_cost + config.cost; | ||
| end | ||
| test_cost = test_cost / (size(test_samples, 4) / config.batch_size); | ||
| fprintf('\nepoc %d, training avg cost: %f, test avg cost: %f\n', epoc, cost_avg, test_cost); | ||
|
|
||
| save_weights(strcat('applications/Shepard_CNN/Shepard_super_res/results/shepard_layer_x2/epoc', num2str(epoc), '.mat')); | ||
|
|
||
| cost_avg = 0; | ||
| end | ||
| end | ||
| end | ||
| end | ||
|
|
Oops, something went wrong.