fixed noise error.

STVIR · Jul 5, 2019 · 34877bb · 34877bb
1 parent bae9b10
commit 34877bb
Show file tree

Hide file tree

Showing 2 changed files with 36 additions and 22 deletions.
diff --git a/src/dataset.py b/src/dataset.py
@@ -1,5 +1,5 @@
 from datasets.dataset_path import *
-
+import os
 
 def get_training_set(opt):
     assert opt.dataset in ['cityscapes', 'cityscapes_two_path', 'kth']
@@ -67,8 +67,8 @@ def get_test_set(opt):
                              returnpath=True)
 
     elif opt.dataset == 'ucf101':
-        from datasets.ucf101_dataset import UCF101
-        test_Dataset = UCF101(datapath=os.path.join(UCF_101_DATA_PATH, category),
+        from datasets.ucf_dataset import UCF101
+        test_Dataset = UCF101(datapath=os.path.join(UCF_101_DATA_PATH, opt.category),
                               datalist=os.path.join(UCF_101_DATA_PATH, 'list/test%s.txt' % (opt.category.lower())),
                               returnpath=True)
 

diff --git a/src/test_refine.py b/src/test_refine.py
@@ -13,7 +13,7 @@
 from opts import parse_opts
 
 args = parse_opts()
-print (args)
+print(args)
 
 
 def make_save_dir(output_image_dir):
@@ -50,8 +50,10 @@ def __init__(self, opt):
 
         test_Dataset = get_test_set(opt)
 
-        self.sampledir = os.path.join('../city_scapes_test_results', self.jobname,
-                                      self.suffix + '_' + str(self.iter_to_load)+'_'+str(opt.seed))
+        # self.sampledir = os.path.join('../city_scapes_test_results', self.jobname,
+        #                               self.suffix + '_' + str(self.iter_to_load)+'_'+str(opt.seed))
+
+        self.sampledir = os.path.join('../ucf101_results', opt.category, self.suffix + '_' + str(opt.seed))
 
         if not os.path.exists(self.sampledir):
             os.makedirs(self.sampledir)
@@ -82,22 +84,31 @@ def test(self):
         opt = self.opt
 
         gpu_ids = range(torch.cuda.device_count())
-        print ('Number of GPUs in use {}'.format(gpu_ids))
+        print('Number of GPUs in use {}'.format(gpu_ids))
 
         iteration = 0
 
         if torch.cuda.device_count() > 1:
-            vae = nn.DataParallel(VAE(hallucination=self.useHallucination, opt=opt, refine=self.refine), device_ids=gpu_ids).cuda()
+            vae = nn.DataParallel(VAE(hallucination=self.useHallucination, opt=opt, refine=self.refine),
+                                  device_ids=gpu_ids).cuda()
         else:
             vae = VAE(hallucination=self.useHallucination, opt=opt, refine=self.refine).cuda()
 
         print(self.jobname)
 
         if self.load:
             # model_name = '../' + self.jobname + '/{:06d}_model.pth.tar'.format(self.iter_to_load)
-            model_name = '../pretrained_models/cityscapes/refine_genmask_098000.pth.tar'
 
-            print ("loading model from {}".format(model_name))
+            if opt.dataset == 'cityscapes':
+                model_name = '../pretrained_models/cityscapes/refine_genmask_098000.pth.tar'
+
+            elif opt.dataset == 'ucf101':
+                model_name = '../pretrained_models/' + opt.dataset + '/' + opt.category.lower() + '_model.pth.tar'
+
+            else:
+                model_name = '../pretrained_models/' + opt.dataset + '/' + opt.category + '_16frames_model.pth.tar'
+
+            print("loading model from {}".format(model_name))
 
             state_dict = torch.load(model_name)
             if torch.cuda.device_count() > 1:
@@ -106,8 +117,7 @@ def test(self):
                 vae.load_state_dict(state_dict['vae'])
 
         z_noise = torch.ones(1, 1024).normal_()
-        for sample,_, paths in tqdm(iter(self.testloader)):
-
+        for sample, _, paths in tqdm(iter(self.testloader)):
             # Set to evaluation mode (randomly sample z from the whole distribution)
             vae.eval()
 
@@ -119,29 +129,33 @@ def test(self):
             # data = data.repeat(1, opt.num_frames, 1, 1, 1)
 
             frame1 = data[:, 0, :, :, :]
+
             noise_bg = Vb(torch.randn(frame1.size())).cuda()
-            z_m = Vb(z_noise.repeat(frame1.size()[0] * 8, 1))
 
-            y_pred_before_refine, y_pred, mu, logvar, flow, flowback, mask_fw, mask_bw = vae(frame1, data, noise_bg, z_m)
+            z_m = Vb(z_noise.repeat(frame1.size()[0] * 4 * int(frame1.shape[-1] / 128), 1))
 
-            utils.save_samples(data, y_pred_before_refine, y_pred, flow, mask_fw, mask_bw, iteration, self.sampledir, opt,
-                         eval=True, useMask=True)
+            y_pred_before_refine, y_pred, mu, logvar, flow, flowback, mask_fw, mask_bw = vae(frame1, data, noise_bg,
+                                                                                             z_m)
+
+            utils.save_samples(data, y_pred_before_refine, y_pred, flow, mask_fw, mask_bw, iteration, self.sampledir,
+                               opt,
+                               eval=True, useMask=True)
 
             utils.save_images(self.output_image_dir, data, y_pred, paths, opt)
             utils.save_images(self.output_image_before_dir, data, y_pred_before_refine, paths, opt)
 
             data = data.cpu().data.transpose(2, 3).transpose(3, 4).numpy()
             utils.save_gif(data * 255, opt.num_frames, [8, 4], self.sampledir + '/{:06d}_real.gif'.format(iteration))
 
-            utils.save_flows(self.output_fw_flow_dir, flow, paths)
-            utils.save_flows(self.output_bw_flow_dir, flowback, paths)
-
-            utils.save_occ_map(self.output_fw_mask_dir, mask_fw, paths)
-            utils.save_occ_map(self.output_bw_mask_dir, mask_bw, paths)
+            # utils.save_flows(self.output_fw_flow_dir, flow, paths)
+            # utils.save_flows(self.output_bw_flow_dir, flowback, paths)
+            #
+            # utils.save_occ_map(self.output_fw_mask_dir, mask_fw, paths)
+            # utils.save_occ_map(self.output_bw_mask_dir, mask_bw, paths)
 
             iteration += 1
 
 
 if __name__ == '__main__':
     a = flowgen(opt=args)
-    a.test()
+    a.test()