fine tune

aaa_helpers.txt

@@ -1,3 +1,10 @@
+nohup python cvae_pyramid.py --name="lhunor_pyr_abl_32_64_e2e_r3" > nohup/review/pyr_abl_32_64_e2e_r3.out 2>&1 &
+nohup python cvae_pyramid.py --name="lhunor_pyr_abl_32_64_e2e_r3_pre" > nohup/review/pyr_abl_32_64_e2e_r3_pre.out 2>&1 &
+nohup python cvae_pyramid.py --name="lhunor_pyr_abl_32_64_e2e_r3_fine" > nohup/review/pyr_abl_32_64_e2e_r3_fine.out 2>&1 &
+nohup python evaluate_ae_pyramid.py --name evaluate_pyr_abl_32_64_e2e_r3 > nohup/evaluate/pyr_abl_32_64_e2e_r3.out 2>&1 &
+nohup python calculate_error.py --name error_pyr_abl_32_64_e2e_r3 > nohup/error/pyr_abl_32_64_e2e_r3.out 2>&1 &
+
+
 one final convolutional layer for smoothing, on the final coordinates
 
 skirts!!!!

calculate_error.py

@@ -1,6 +1,6 @@
 import os
 import sys
-from configs.config_abl_all_post import config
+from configs.config_abl_pyr_32 import config
 
 os.environ["CUDA_VISIBLE_DEVICES"] = config['gpu']
 # import tensorflow as tf
@@ -59,13 +59,13 @@ if __name__ == '__main__':
     folder = config['test_dir']
 
     checkpoint_name = config['name']
-    iteration = ''
+    iteration = '200000'
 
     F = np.load("custom_utils/smpl_faces" + skirt_modifier + ".npy").astype(np.int32)
     garment_type_table = np.load('custom_data/garment_type_table_test.npy', allow_pickle=True).item()
 
 
-    output_folder = '/home/workspace/data/results/post/' + checkpoint_name + '-' + iteration
+    output_folder = '/home/workspace/data/results/' + checkpoint_name + '-' + iteration
 
 
     faces_folder = '/home/workspace/data/results/support/faces' + ('_skirt' if config['is_skirt'] else '')

configs/config_abl_pyr_32.py

@@ -1,8 +1,9 @@
 config = dict()
 
 config['gpu'] = '0'
-config['name'] = 'pyr_abl_32'
+config['name'] = 'pyr_abl_32_64_e2e_r3_fine_2'
 config['notes'] = ''
+config['tags'] = ['review_3']
 
 # DON'T FORGET TO RESET SSIM LOSS USAGE
 # DON'T FORGET TO CHECK MODEL LOADING
@@ -21,9 +22,12 @@ config['use_offsets'] =               False
 
 
 config['pyramid'] = dict()
-config['pyramid']['resolutions'] = [32]
+config['pyramid']['resolutions'] = [32, 64]
 config['pyramid']['pretrained_pyramid'] = ''
-config['pyramid']['final_upscale_enhancer'] = 'checkpoints/32_512_4'
+config['pyramid']['final_upscale_enhancer'] = 'checkpoints/64_512_2'
+
+config['pretrained_model'] = 'checkpoints/pyr_abl_32_64_e2e_r3_fine'
+config['pretrained_model_iteration'] = ''
 
 config['train_dir'] = '/home/workspace/local-data/train'
 config['valid_dir'] = '/home/workspace/local-data/valid'
@@ -40,7 +44,7 @@ config['smpl_res'] = 14475
 if config['is_cape']:
   config['smpl_res'] = 3638
 
-config['lr'] = 0.0001
+config['lr'] = 0.00001
 config['starting_balancing_coeff'] = 0.0001
 config['reg_loss_coeff'] = 0.1 # set to 0 if don't want to use edge+collision
 config['avg_garment_path'] = ''
@@ -64,7 +68,7 @@ config['pyramid']['latent_dim'] = config['latent_dim']
 config['pyramid']['add_noise'] = True
 config['pyramid']['noise_amount'] = 0.025
 
-config['pyramid']['loss_weights'] = [1,1,1,1,1]
+config['pyramid']['loss_weights'] = [2,1,1,1,1]
 
 for res in config['pyramid']['resolutions']:
   if res != 512:
@@ -149,7 +153,7 @@ for res in config['pyramid']['resolutions']:
     level['decoder_depths'] = [9, 9]
     level['decoder_dims'] = [256, 256]
     level['avg_garment_path'] = ''
-    level['upscale_enhancer'] = ''#'checkpoints/up_3_d32_64_2'
+    level['upscale_enhancer'] = 'checkpoints/32_512_4'#'checkpoints/up_3_d32_64_2'
     level['upscale_enhancer_freeze'] = True
     config['pyramid']['levels'].append(level)
   else:

configs/config_abl_pyr_32_skirt.py

@@ -1,8 +1,8 @@
 config = dict()
 
 config['gpu'] = '3'
-config['name'] = 'pyr_abl_32_skirt_2'
+config['name'] = 'pyr_abl_32_skirt_r1'
-config['notes'] = 'continued from pyr_abl_32_skirt-180k'
+config['notes'] = ''
 
 # DON'T FORGET TO RESET SSIM LOSS USAGE
 # DON'T FORGET TO CHECK MODEL LOADING
@@ -24,7 +24,10 @@ config['pyramid']['resolutions'] = [32]
 config['pyramid']['pretrained_pyramid'] = ''
 config['pyramid']['final_upscale_enhancer'] = 'checkpoints/32_512_skirt'
 
-config['train_dir'] = '/home/workspace/local-data/train_skirt'
+config['pretrained_model'] = ''
+config['pretrained_model_iteration'] = ''
+
+config['train_dir'] = '/home/workspace/data/dataset/local-cache/train_skirt'
 config['valid_dir'] = '/home/workspace/local-data/valid_skirt'
 config['test_dir'] = '/home/workspace/data/dataset/test_skirt_ready_2'
 config['batch_size'] = 4
@@ -35,7 +38,7 @@ config['pyramid']['static_model'] = 'checkpoints/static_1_skirt'
 config['pyramid']['body_model'] = 'checkpoints/body_1_skirt_2_2'
 config['pyramid']['curvature_model'] = 'checkpoints/curv_1_skirt'
 
-config['dataset_in_memory'] = False
+config['dataset_in_memory'] = True
 config['resolution'] = None
 config['latent_dim'] = 128
 
@@ -65,6 +68,7 @@ config['signature_static'] = { 'uv_static': (512, 256, 3),
 config['pyramid']['is_label'] = False
 config['pyramid']['latent_dim'] = config['latent_dim']
 config['pyramid']['add_noise'] = True
+config['pyramid']['noise_amount'] = 0.001
 
 config['pyramid']['loss_weights'] = [1,1,1,1,1]
 

custom_utils/custom_utils.py

@@ -45,7 +45,7 @@ class UV2Mesh(tf.keras.Model):
         if uv.shape[1] == 3:
             uv = tf.transpose(uv, [0,2,3,1])
         if uv.shape[1] != 512:
-            uv = tf.image.resize(uv, (512, 256), tf.image.ResizeMethod.LANCZOS5)
+            uv = tf.image.resize(uv, (512, 256), tf.image.ResizeMethod.BILINEAR)
             
         batch_size = uv.shape[0]
 

cvae_pyramid.py

@@ -1,6 +1,6 @@
 import os
 import sys
-from configs.post.config_abl_pyr_64_128_5 import config
+from configs.config_abl_pyr_32 import config
 
 os.environ["CUDA_VISIBLE_DEVICES"] = config['gpu']
 import tensorflow as tf
@@ -41,7 +41,7 @@ from losses import collision_loss as collision_penalty
 from custom_utils.custom_utils import batch_to_dict, load_model
 from dataset import DatasetFactory, StaticDataset
 from custom_utils.smpl import SMPL
-from models.pyramid_model_optimized import PyramidNew as Pyramid
+from models.pyramid_model_optimized import Pyramid
 from postprocessing import fix_collision
 
 
@@ -59,9 +59,11 @@ def init_wandb():
     artifact.add_file('models/cond_encoder.py')
     artifact.add_file('dataset.py')
     artifact.add_file('convnext/models/convnext_tf.py')
-    config_file_names = [key for key in sys.modules.keys()  if key.startswith('configs.post.')]
+    config_file_names = [key for key in sys.modules.keys()  if key.startswith('configs')]
     for config_file_name in config_file_names:
-        artifact.add_file('/'.join(config_file_name.split('.')) + '.py')
+        file_name = '/'.join(config_file_name.split('.')) + '.py'
+        if os.path.exists(file_name):
+            artifact.add_file(file_name)
     run.log_artifact(artifact)
 
 uv2mesh = UV2Mesh(res=256, is_skirt=config['is_skirt'], is_smpl_orig=config['is_cape'])
@@ -215,66 +217,70 @@ def train_step(batch):
     with tf.GradientTape() as tape:
         out_levels = model([levels_input, conds], training=True)
 
-        if config['use_skinning']:
+        mesh_orig, uv_orig = batch['mesh_unposed'], batch['uv_unposed']
-            mesh_orig, uv_orig = batch['mesh_unposed'], batch['uv_unposed']
-        else:
-            mesh_orig, uv_orig = batch['mesh_posed'], batch['uv_posed']
 
         output = dict()
         reconstr_loss_total = 0
         kl_total = 0
 
-        #for i, res in enumerate(config['pyramid']['resolutions']):
+        for i, res in enumerate(config['pyramid']['resolutions']):
-        i = -1
+            # i = -1
-        res = config['pyramid']['resolutions'][-1]
+            # res = config['pyramid']['resolutions'][-1]
-        res_modifier = '_' + str(res) if res != 512 else ''
+            if res != 512:
-        pred, mean, logvar = out_levels[i]
+                if i == len(config['pyramid']['resolutions']) - 1:
+                    res_modifier = ''
+                else:
+                    res_modifier = '_' + str(res)
+            else:
+                res_modifier = ''
 
-        mesh_pred = uv2mesh(pred)
+            pred, mean, logvar = out_levels[i]
 
-        if config['use_skinning']:
+            mesh_pred = uv2mesh(pred)
-            shape = batch['pose_shape'][:, 72:]
-            gender = tf.cast(tf.greater_equal(shape[:, 10] , 0.5), tf.int32)
-            shape_offsets = tf.stack([tf.gather(tf.tensordot(shape[i,:10][tf.newaxis, :], smpl.shapedirs[gender[i]], axes=[[1], [2]])[0,:,:], smpl.to_keep, axis=0) for i in range(shape.shape[0])])
 
-            mesh_pred_shaped = mesh_pred + shape_offsets * batch['mesh_mask'][:,:,tf.newaxis]
+            if config['use_skinning']:
-            mesh_pred_posed = smpl.run_garment(batch['pose_shape'][:, :72], batch['pose_shape'][:, 72:], mesh_pred_shaped)
+                shape = batch['pose_shape'][:, 72:]
-        else:
+                gender = tf.cast(tf.greater_equal(shape[:, 10] , 0.5), tf.int32)
-            mesh_pred_posed = mesh_pred
+                shape_offsets = tf.stack([tf.gather(tf.tensordot(shape[i,:10][tf.newaxis, :], smpl.shapedirs[gender[i]], axes=[[1], [2]])[0,:,:], smpl.to_keep, axis=0) for i in range(shape.shape[0])])
 
-        # is the output already masked?
+                mesh_pred_shaped = mesh_pred + shape_offsets * batch['mesh_mask'][:,:,tf.newaxis]
+                mesh_pred_posed = smpl.run_garment(batch['pose_shape'][:, :72], batch['pose_shape'][:, 72:], mesh_pred_shaped)
+            else:
+                mesh_pred_posed = mesh_pred
 
-        e_loss = tf.reduce_mean(edge_loss(mesh_pred, mesh_orig, batch['mesh_mask']))
+            # is the output already masked?
-        c_loss, c_ratio = collision_penalty(mesh_pred_posed, batch['mesh_body_posed'], batch['mesh_mask'])
-        c_loss, c_ratio = tf.reduce_mean(c_loss), tf.reduce_mean(c_ratio)
-        l2_loss = tf.reduce_mean(1e3 * tf.reduce_mean(tf.reduce_sum(((uv_orig - pred) ** 2), axis=-1), axis=[1, 2]))
-        # ms_ssim = tf.reduce_mean(tf.image.ssim_multiscale(pred + 5, batch['uv_unposed'] + 5, max_val=10, power_factors=(0.5, 0.2, 0.1, 0.1, 0.1)))
-        # ms_ssim = tf.reduce_mean(tf.image.ssim(pred + 1, batch['uv_unposed'] + 1, max_val=3))
-        l2_mesh_loss = tf.reduce_mean(1e3 * tf.reduce_mean(tf.reduce_sum(batch['mesh_mask'][:,:,tf.newaxis] * (batch['mesh_posed'] - mesh_pred_posed) ** 2, axis=-1), axis=-1))
-        n_loss = normal_loss(mesh_pred_posed, batch['mesh_posed'], batch['mesh_mask'], is_skirt=config['is_skirt'])
-        kl = kl_loss(mean, logvar)
 
-        if config['use_extra_losses']:
+            e_loss = tf.reduce_mean(edge_loss(mesh_pred, mesh_orig, batch['mesh_mask']))
-            reconstr_loss_total = reconstr_loss_total + config['pyramid']['loss_weights'][i] * (l2_loss + l2_mesh_loss)
+            c_loss, c_ratio = collision_penalty(mesh_pred_posed, batch['mesh_body_posed'], batch['mesh_mask'])
-            if not config['is_cape']:
+            c_loss, c_ratio = tf.reduce_mean(c_loss), tf.reduce_mean(c_ratio)
-                reconstr_loss_total = reconstr_loss_total + config['pyramid']['loss_weights'][i] * (e_loss + c_loss) * config['reg_loss_coeff'] 
+            l2_loss = tf.reduce_mean(1e3 * tf.reduce_mean(tf.reduce_sum(((batch['uv_unposed' + res_modifier] - pred) ** 2), axis=-1), axis=[1, 2]))
-        else:
+            # ms_ssim = tf.reduce_mean(tf.image.ssim_multiscale(pred + 5, batch['uv_unposed'] + 5, max_val=10, power_factors=(0.5, 0.2, 0.1, 0.1, 0.1)))
-            reconstr_loss_total = reconstr_loss_total + config['pyramid']['loss_weights'][i] * (l2_loss)
+            # ms_ssim = tf.reduce_mean(tf.image.ssim(pred + 1, batch['uv_unposed'] + 1, max_val=3))
+            l2_mesh_loss = tf.reduce_mean(1e3 * tf.reduce_mean(tf.reduce_sum(batch['mesh_mask'][:,:,tf.newaxis] * (batch['mesh_posed'] - mesh_pred_posed) ** 2, axis=-1), axis=-1))
+            n_loss = normal_loss(mesh_pred_posed, batch['mesh_posed'], batch['mesh_mask'], is_skirt=config['is_skirt'])
+            kl = kl_loss(mean, logvar)
 
-        if config['use_norm_loss'] and step > 2000:
+            if config['use_extra_losses']:
-            reconstr_loss_total = reconstr_loss_total + n_loss * config['reg_loss_coeff']
+                reconstr_loss_total = reconstr_loss_total + config['pyramid']['loss_weights'][i] * (l2_loss + l2_mesh_loss)
+                if not config['is_cape']:
+                    reconstr_loss_total = reconstr_loss_total + config['pyramid']['loss_weights'][i] * (e_loss + c_loss) * config['reg_loss_coeff'] 
+            else:
+                reconstr_loss_total = reconstr_loss_total + config['pyramid']['loss_weights'][i] * (l2_loss)
 
-        kl_total += kl
+            # if config['use_norm_loss'] and step > 2000:
+            #     reconstr_loss_total = reconstr_loss_total + n_loss * config['reg_loss_coeff']
             
-        mean = tf.reduce_mean(mean)
+            kl_total += kl
-        logvar = tf.reduce_mean(logvar)
 
-        metric = tf.reduce_sum(tf.math.sqrt(tf.reduce_sum(batch['mesh_mask'][:,:,tf.newaxis] * (mesh_orig - mesh_pred) ** 2, axis=-1))) / tf.reduce_sum(batch['mesh_mask'])
+            mean = tf.reduce_mean(mean)
+            logvar = tf.reduce_mean(logvar)
 
-        output = { **output, **{ 'train/edge_loss' + res_modifier: e_loss, 'train/collision_loss' + res_modifier: c_loss,'train/euclid_error' + res_modifier: metric, 
+            metric = tf.reduce_sum(tf.math.sqrt(tf.reduce_sum(batch['mesh_mask'][:,:,tf.newaxis] * (mesh_orig - mesh_pred) ** 2, axis=-1))) / tf.reduce_sum(batch['mesh_mask'])
-                                 'train/mean' + res_modifier: mean, 'train/logvar' + res_modifier: logvar, 'train/kl_loss' + res_modifier: kl,
+
-                                 'train/var' + res_modifier: np.e ** logvar, 'train/l2_loss' + res_modifier: l2_loss, 'train/normal_loss': n_loss,
+            output = { **output, **{ 'train/edge_loss' + res_modifier: e_loss, 'train/collision_loss' + res_modifier: c_loss,'train/euclid_error' + res_modifier: metric, 
-                                 'train/euclid_error': metric, 'train/l2_mesh_loss': l2_mesh_loss } } # 'train/ms_ssim' + res_modifier: ms_ssim,
+                                    'train/mean' + res_modifier: mean, 'train/logvar' + res_modifier: logvar, 'train/kl_loss' + res_modifier: kl,
+                                    'train/var' + res_modifier: np.e ** logvar, 'train/l2_loss' + res_modifier: l2_loss, 'train/normal_loss': n_loss,
+                                    'train/euclid_error': metric, 'train/l2_mesh_loss': l2_mesh_loss } } # 'train/ms_ssim' + res_modifier: ms_ssim,
 
         loss = reconstr_loss_total + balancing_coeff * kl_total
 
@@ -288,66 +294,69 @@ def valid_step(batch):
     levels_input, conds = group_inputs_for_levels(batch, config)
     out_levels = model([levels_input, conds], training=False)
 
-
+    mesh_orig, uv_orig = batch['mesh_unposed'], batch['uv_unposed']
-    if config['use_skinning']:
-        mesh_orig, uv_orig = batch['mesh_unposed'], batch['uv_unposed']
-    else:
-        mesh_orig, uv_orig = batch['mesh_posed'], batch['uv_posed']
 
     output = dict()
     reconstr_loss_total = 0
     kl_total = 0
 
-    #for i, res in enumerate(config['pyramid']['resolutions']):
+    for i, res in enumerate(config['pyramid']['resolutions']):
-    i = -1
+        # i = -1
-    res = config['pyramid']['resolutions'][-1]
+        # res = config['pyramid']['resolutions'][-1]
-    res_modifier = '_' + str(res) if res != 512 else ''
+        if res != 512:
-    pred, mean, logvar = out_levels[i]
+            if i == len(config['pyramid']['resolutions']) - 1:
+                res_modifier = ''
+            else:
+                res_modifier = '_' + str(res)
+        else:
+            res_modifier = ''
 
-    mesh_pred = uv2mesh(pred)
+        pred, mean, logvar = out_levels[i]
 
-    if config['use_skinning']:
+        mesh_pred = uv2mesh(pred)
-        shape = batch['pose_shape'][:, 72:]
-        gender = tf.cast(tf.greater_equal(shape[:, 10] , 0.5), tf.int32)
-        shape_offsets = tf.stack([tf.gather(tf.tensordot(shape[i,:10][tf.newaxis, :], smpl.shapedirs[gender[i]], axes=[[1], [2]])[0,:,:], smpl.to_keep, axis=0) for i in range(shape.shape[0])])
 
-        mesh_pred_shaped = mesh_pred + shape_offsets * batch['mesh_mask'][:,:,tf.newaxis]
+        if config['use_skinning']:
-        mesh_pred_posed = smpl.run_garment(batch['pose_shape'][:, :72], batch['pose_shape'][:, 72:], mesh_pred_shaped)
+            shape = batch['pose_shape'][:, 72:]
-    else:
+            gender = tf.cast(tf.greater_equal(shape[:, 10] , 0.5), tf.int32)
-        mesh_pred_posed = mesh_pred
+            shape_offsets = tf.stack([tf.gather(tf.tensordot(shape[i,:10][tf.newaxis, :], smpl.shapedirs[gender[i]], axes=[[1], [2]])[0,:,:], smpl.to_keep, axis=0) for i in range(shape.shape[0])])
-    # is the output already masked?
 
-    e_loss = tf.reduce_mean(edge_loss(mesh_pred, mesh_orig, batch['mesh_mask']))
+            mesh_pred_shaped = mesh_pred + shape_offsets * batch['mesh_mask'][:,:,tf.newaxis]
-    c_loss, c_ratio = collision_penalty(mesh_pred_posed, batch['mesh_body_posed'], batch['mesh_mask'])
+            mesh_pred_posed = smpl.run_garment(batch['pose_shape'][:, :72], batch['pose_shape'][:, 72:], mesh_pred_shaped)
-    c_loss, c_ratio = tf.reduce_mean(c_loss), tf.reduce_mean(c_ratio)
+        else:
-    l2_loss = tf.reduce_mean(1e3 * tf.reduce_mean(tf.reduce_sum(((uv_orig - pred) ** 2), axis=-1), axis=[1, 2]))
+            mesh_pred_posed = mesh_pred
-    # ms_ssim = tf.reduce_mean(tf.image.ssim_multiscale(pred + 5, batch['uv_unposed'] + 5, max_val=10, power_factors=(0.5, 0.2, 0.1, 0.1, 0.1)))
+        # is the output already masked?
-    # ms_ssim = tf.reduce_mean(tf.image.ssim(pred + 1, batch['uv_unposed'] + 1, max_val=3))
-    l2_mesh_loss = tf.reduce_mean(1e3 * tf.reduce_mean(tf.reduce_sum(batch['mesh_mask'][:,:,tf.newaxis] * (batch['mesh_posed'] - mesh_pred_posed) ** 2, axis=-1), axis=-1))
-    n_loss = normal_loss(mesh_pred_posed, batch['mesh_posed'], batch['mesh_mask'], is_skirt=config['is_skirt'])
-    kl = kl_loss(mean, logvar)
 
-    if config['use_extra_losses']:
+        e_loss = tf.reduce_mean(edge_loss(mesh_pred, mesh_orig, batch['mesh_mask']))
-        reconstr_loss_total = reconstr_loss_total + config['pyramid']['loss_weights'][i] * (l2_loss + l2_mesh_loss)
+        c_loss, c_ratio = collision_penalty(mesh_pred_posed, batch['mesh_body_posed'], batch['mesh_mask'])
-        if not config['is_cape']:
+        c_loss, c_ratio = tf.reduce_mean(c_loss), tf.reduce_mean(c_ratio)
-            reconstr_loss_total = reconstr_loss_total + config['pyramid']['loss_weights'][i] * (e_loss + c_loss) * config['reg_loss_coeff'] 
+        l2_loss = tf.reduce_mean(1e3 * tf.reduce_mean(tf.reduce_sum(((batch['uv_unposed' + res_modifier] - pred) ** 2), axis=-1), axis=[1, 2]))
-    else:
+        # ms_ssim = tf.reduce_mean(tf.image.ssim_multiscale(pred + 5, batch['uv_unposed'] + 5, max_val=10, power_factors=(0.5, 0.2, 0.1, 0.1, 0.1)))
-        reconstr_loss_total = reconstr_loss_total + config['pyramid']['loss_weights'][i] * (l2_loss)
+        # ms_ssim = tf.reduce_mean(tf.image.ssim(pred + 1, batch['uv_unposed'] + 1, max_val=3))
+        l2_mesh_loss = tf.reduce_mean(1e3 * tf.reduce_mean(tf.reduce_sum(batch['mesh_mask'][:,:,tf.newaxis] * (batch['mesh_posed'] - mesh_pred_posed) ** 2, axis=-1), axis=-1))
+        n_loss = normal_loss(mesh_pred_posed, batch['mesh_posed'], batch['mesh_mask'], is_skirt=config['is_skirt'])
+        kl = kl_loss(mean, logvar)
 
-    if config['use_norm_loss'] and step > 2000:
+        if config['use_extra_losses']:
-        reconstr_loss_total = reconstr_loss_total + n_loss * config['reg_loss_coeff']
+            reconstr_loss_total = reconstr_loss_total + config['pyramid']['loss_weights'][i] * (l2_loss + l2_mesh_loss)
+            if not config['is_cape']:
+                reconstr_loss_total = reconstr_loss_total + config['pyramid']['loss_weights'][i] * (e_loss + c_loss) * config['reg_loss_coeff'] 
+        else:
+            reconstr_loss_total = reconstr_loss_total + config['pyramid']['loss_weights'][i] * (l2_loss)
         
-    kl_total += kl
+        # if config['use_norm_loss'] and step > 2000:
+        #     reconstr_loss_total = reconstr_loss_total + n_loss * config['reg_loss_coeff']
             
-    mean = tf.reduce_mean(mean)
+        kl_total += kl
-    logvar = tf.reduce_mean(logvar)
 
-    metric = tf.reduce_sum(tf.math.sqrt(tf.reduce_sum(batch['mesh_mask'][:,:,tf.newaxis] * (mesh_orig - mesh_pred) ** 2, axis=-1))) / tf.reduce_sum(batch['mesh_mask'])
+        mean = tf.reduce_mean(mean)
+        logvar = tf.reduce_mean(logvar)
 
-    output = { **output, **{ 'valid/edge_loss' + res_modifier: e_loss, 'valid/collision_loss' + res_modifier: c_loss,'valid/euclid_error' + res_modifier: metric, 
+        metric = tf.reduce_sum(tf.math.sqrt(tf.reduce_sum(batch['mesh_mask'][:,:,tf.newaxis] * (mesh_orig - mesh_pred) ** 2, axis=-1))) / tf.reduce_sum(batch['mesh_mask'])
-                                 'valid/mean' + res_modifier: mean, 'valid/logvar' + res_modifier: logvar, 'valid/kl_loss' + res_modifier: kl,
+
-                                'valid/var' + res_modifier: np.e ** logvar, 'valid/l2_loss' + res_modifier: l2_loss, 'valid/normal_loss': n_loss,
+        output = { **output, **{ 'valid/edge_loss' + res_modifier: e_loss, 'valid/collision_loss' + res_modifier: c_loss,'valid/euclid_error' + res_modifier: metric, 
-                                'valid/euclid_error': metric, 'valid/l2_mesh_loss': l2_mesh_loss } } # 'valid/ms_ssim' + res_modifier: ms_ssim,
+                                    'valid/mean' + res_modifier: mean, 'valid/logvar' + res_modifier: logvar, 'valid/kl_loss' + res_modifier: kl,
+                                    'valid/var' + res_modifier: np.e ** logvar, 'valid/l2_loss' + res_modifier: l2_loss, 'valid/normal_loss': n_loss,
+                                    'valid/euclid_error': metric, 'valid/l2_mesh_loss': l2_mesh_loss } } # 'valid/ms_ssim' + res_modifier: ms_ssim,
 
     loss = reconstr_loss_total + balancing_coeff * kl_total
 
@@ -790,7 +799,7 @@ if __name__ == '__main__':
     # generate_results()
 
     ckpt = tf.train.Checkpoint(model=model)
-    manager = tf.train.CheckpointManager(ckpt, 'checkpoints/post/' + config['name'], max_to_keep=20)
+    manager = tf.train.CheckpointManager(ckpt, 'checkpoints/' + config['name'], max_to_keep=20)
 
     # load_model(model, 'checkpoints/pyr_abl_32_skirt', iteration='180000')
 
@@ -798,30 +807,33 @@ if __name__ == '__main__':
         # assert len(config['pyramid']['resolutions']) == 1
         load_model(model, config['pretrained_model'], iteration=config['pretrained_model_iteration'])
 
-    if len(config['pyramid']['resolutions']) == 1:
+    model.static_enc.trainable = False
-        model.static_enc.trainable = False
+    model.body_enc.trainable = False
-        model.body_enc.trainable = False
+    model.curv_enc.trainable = False
-        model.curv_enc.trainable = False
-    else:
-        load_model(model, config['pyramid']['pretrained_pyramid'], iteration=(config['pyramid']['pretrained_pyramid_iteration'] if 'pretrained_pyramid_iteration' in config.keys() else ''))
-        model.static_enc.trainable = False
-        model.body_enc.trainable = False
-        model.curv_enc.trainable = False
 
+    # model.trainable = False
+    # model.cond_encoder_2.trainable = True
+    # if len(config['pyramid']['resolutions']) == 1:
+    # else:
+    #     load_model(model, config['pyramid']['pretrained_pyramid'], iteration=(config['pyramid']['pretrained_pyramid_iteration'] if 'pretrained_pyramid_iteration' in config.keys() else ''))
+    #     model.static_enc.trainable = False
+    #     model.body_enc.trainable = False
+    #     model.curv_enc.trainable = False
+    #     model.cond_encoder.trainable = False
 
     # load_model(model.levels[0].upscale_enhancer, config['pyramid']['levels'][0]['upscale_enhancer'])
     # model.levels[0].upscale_enhancer.trainable = False
 
 
-    for i in range(len(model.levels)):
+    for i in range(len(model.levels)-1):
         load_model(model.levels[i].upscale_enhancer, config['pyramid']['levels'][i]['upscale_enhancer'])
         model.levels[i].upscale_enhancer.trainable = False
-    # if config['pyramid']['resolutions'][-1] != 512:
+    if config['pyramid']['resolutions'][-1] != 512:
-    #     load_model(model.upscale_enhancer, config['pyramid']['final_upscale_enhancer'])
+        load_model(model.upscale_enhancer, config['pyramid']['final_upscale_enhancer'])
-    #     model.upscale_enhancer.trainable = False
+        model.upscale_enhancer.trainable = False
 
-    for i in range(len(model.levels)-1):
+    # for i in range(len(model.levels)-1):
-        model.levels[i].trainable = False
+    #     model.levels[i].trainable = False
 
     # load_model(model.cond_encoder, config['pyramid']['pretrained_pyramid'])
     # load_model(model.levels[0], config['pyramid']['pretrained_pyramid'])

evaluate_ae_pyramid.py

@@ -1,6 +1,6 @@
 import os
 import sys
-from configs.config_abl_2_pyr_32_64_128_256_512 import config
+from configs.config_abl_pyr_32 import config
 
 os.environ["CUDA_VISIBLE_DEVICES"] = config['gpu']
 import tensorflow as tf
@@ -97,7 +97,7 @@ if __name__ == '__main__':
     ids = [id.split('_')[0] for id in ids]
     ids = set(ids)
 
-    iteration = '50000'
+    iteration = '200000'
 
     load_model(model, 'checkpoints/' + checkpoint_name, iteration=iteration)
     # load_model(model, 'checkpoints/' + 'abl_all_120k')

models/cvae_model.py

@@ -43,7 +43,7 @@ class CVAE(tf.keras.Model):
         self.cond_encoder = tf.keras.Sequential([tf.keras.layers.Dense(512, activation='relu'),
                                                  tf.keras.layers.Dense(256, activation='relu'),
                                                  tf.keras.layers.Dense(128, activation='relu'),
-                                                 tf.keras.layers.Lambda(lambda x: x / 1000),
+                                                #  tf.keras.layers.Lambda(lambda x: x / 1000),
                                                  tf.keras.layers.Dense(config['latent_dim'], activation='tanh')])
 
         self.encoder = VariationalEncoder(config['cvae']['latent_dim'], config['cvae']['encoder_depths'], config['cvae']['encoder_dims'], kernel_size=config['cvae']['kernel_size'])

models/pyramid_level.py

@@ -92,7 +92,7 @@ class PyramidLevel(tf.keras.Model):
         
         # x = x * uv_mask_high[:,:,:,tf.newaxis]
 
-        return x, mean, logvar
+        return x, z, mean, logvar
 
 
 

models/pyramid_model_optimized.py

@@ -78,6 +78,11 @@ class Pyramid(tf.keras.Model):
             load_model(self.upscale_enhancer, config['pyramid']['final_upscale_enhancer'])
             self.upscale_enhancer.trainable = False
 
+        self.cond_encoder_2 = tf.keras.Sequential([tf.keras.layers.Dense(512, activation='relu'),
+                                                 tf.keras.layers.Dense(256, activation='relu'),
+                                                 tf.keras.layers.Dense(128, activation='relu'),
+                                                 tf.keras.layers.Dense(config['pyramid']['latent_dim'], activation='tanh'),])
+
     def enhance(self, low, mask, training):
         x = self.upscale_enhancer([low, mask], training=training)
         return x
@@ -90,44 +95,54 @@ class Pyramid(tf.keras.Model):
         z_cond = tf.concat([z_static, z_body, z_curv], axis=-1)
         z_cond = self.cond_encoder(z_cond, training=training)
 
+        return z_cond, z_static, z_body, z_curv
+
+    def encode_conds_2(self, z, z_static, z_body, z_curv, training):
+        z_cond = tf.concat([z, z_static, z_body, z_curv], axis=-1)
+        z_cond = self.cond_encoder_2(z_cond, training=training)
+
         return z_cond
 
     def call(self, inputs, training):
         levels_input, conds = inputs
         uv_static, uv_body, uv_curv = conds[0], conds[1], conds[2]
 
-        z_cond = self.encode_conds(uv_static, uv_body, uv_curv, training)
+        z_cond, z_static, z_body, z_curv = self.encode_conds(uv_static, uv_body, uv_curv, training)
 
         if self.add_noise == True and training == True:
             z_cond = z_cond + (tf.random.uniform(z_cond.shape, maxval=self.noise_amount * 2) - self.noise_amount)
 
         out_levels = []
-        pred, mean, logvar = self.levels[0](levels_input[0], z_cond, training=training)
+        pred, z, mean, logvar = self.levels[0](levels_input[0], z_cond, training=training)
         if self.use_offsets:
             pred = pred + levels_input[0][2] # uv_body_posed_32
         if 1 != len(self.levels):
             pred = self.levels[0].enhance(pred, levels_input[0][-2], training=training)  # levels_input[i][-2]=uv_mask_res
-            pred = tf.image.resize(pred, (self.levels[1].size, self.levels[1].size//2), tf.image.ResizeMethod.LANCZOS5)
             out_levels.append([pred, mean, logvar])
+            pred = tf.image.resize(pred, (self.levels[1].size, self.levels[1].size//2), tf.image.ResizeMethod.BILINEAR)
             pred = pred * levels_input[1][-2][:,:,:,tf.newaxis]
         else:
             pred = self.enhance(pred, levels_input[0][-2], training=training)
-            pred = tf.image.resize(pred, (512, 256), tf.image.ResizeMethod.LANCZOS5)
+            pred = tf.image.resize(pred, (512, 256), tf.image.ResizeMethod.BILINEAR)
             pred = pred * levels_input[-1][-1][:,:,:,tf.newaxis] # full res uv mask
             out_levels.append([pred, mean, logvar])
 
         for i in range(1, len(self.levels)):
-            pred_level, mean, logvar = self.levels[i](levels_input[i], z_cond, training=training)
+            input = levels_input[i]
+            input[0] = pred # use previous level output as input for next level
+            z_cond_2 = self.encode_conds_2(z, z_static, z_body, z_curv, training=training)
+
+            pred_level, z, mean, logvar = self.levels[i](input, z_cond_2, training=training)
             pred = pred + pred_level
             if i != len(self.levels) - 1:
                 pred = self.levels[i].enhance(pred, levels_input[i][-2], training=training)
                 out_levels.append([pred, mean, logvar])
-                pred = tf.image.resize(pred, (self.levels[i+1].size, self.levels[i+1].size//2), tf.image.ResizeMethod.LANCZOS5)
+                pred = tf.image.resize(pred, (self.levels[i+1].size, self.levels[i+1].size//2), tf.image.ResizeMethod.BILINEAR)
                 pred = pred * levels_input[i+1][-2][:,:,:,tf.newaxis]
             else:
                 if self.upscale_enhancer is not None:
                     pred = self.enhance(pred, levels_input[-1][-2], training=training)
-                    pred = tf.image.resize(pred, (512, 256), tf.image.ResizeMethod.LANCZOS5)
+                    pred = tf.image.resize(pred, (512, 256), tf.image.ResizeMethod.BILINEAR)
                 pred = pred * levels_input[-1][-1][:,:,:,tf.newaxis] # full res uv mask
                 out_levels.append([pred, mean, logvar])