Merge remote-tracking branch 'origin/iccv' into new_model

aaa_helpers.txt

@@ -1,5 +1,9 @@
+nohup python cvae.py --name="lhunor_weights_test_6" > nohup/iccv/weights_test_6.out 2>&1 &
+nohup python evaluate_ae.py --name iccv_weights_test_6 > nohup/iccv/evaluate_iccv_weights_test_6.out 2>&1 &
+nohup python calculate_error.py --name iccv_weights_test_6 > nohup/iccv/error_iccv_weights_test_6.out 2>&1 &
+
+
 one final convolutional layer for smoothing, on the final coordinates
-nohup python calculate_error.py --name error_al_all_150k > nohup/error/post_al_all_150k.out 2>&1 &
 
 skirts!!!!
 pyramid next week

compare_models_saved.py

@@ -0,0 +1,188 @@
+import os
+import sys
+from configs.config_abl_all_post import config as config_a
+
+os.environ["CUDA_VISIBLE_DEVICES"] = '5'
+import tensorflow as tf
+
+gpus = tf.config.experimental.list_physical_devices('GPU')
+if gpus:
+    try:
+        # Currently, memory growth needs to be the same across GPUs
+        for gpu in gpus:
+            tf.config.experimental.set_memory_growth(gpu, True)
+        logical_gpus = tf.config.experimental.list_logical_devices('GPU')
+        print(len(gpus), "Physical GPUs,", len(logical_gpus), "Logical GPUs")
+    except RuntimeError as e:
+        # Memory growth must be set before GPUs have been initialized
+        print(e)
+
+import glob
+import time
+import argparse
+import wandb
+import tqdm
+import random
+import numpy as np
+import plotly.graph_objects as go
+import plotly.express as ex
+from PIL import Image
+from postprocessing import fix_collision
+
+if config_a['is_skirt']:
+    from losses import edge_loss_skirt as edge_loss
+else:
+    from losses import edge_loss
+
+
+
+from losses import collision_ratio, kl_loss #, concept_loss
+# from losses import collision_penalty as collision_penalty
+from losses import collision_loss as collision_penalty
+from custom_utils.custom_utils import batch_to_dict, load_model, plot_3d, UV2Mesh, get_summary_html, maskFacesBatch, _maskFaces, plot_3d_collision
+from dataset import DatasetFactory, StaticDataset
+from custom_utils.smpl import SMPL
+from models.cvae_model import CVAE
+from custom_utils.supersmpl_io import writeOBJ
+from plotly.subplots import make_subplots
+
+
+# uv2mesh = UV2Mesh(res=256, is_skirt=config_a['is_skirt'], is_adaptive=False)
+
+
+
+def plot(mesh_gt, pred_a, pred_b, body, mask):
+    fig = make_subplots(rows=1, cols=3, start_cell="top-left", specs = [ [{ 'type': 'scene' } for _ in range(3)] ])
+    
+    _, faces_garment = _maskFaces(np.zeros((14475, 3), np.float32), F, mask)
+    faces = np.concatenate((F, faces_garment + 14475), axis=0)
+    mesh_gt = np.concatenate((body, mesh_gt[mask, :]), axis=0)
+    pred_a = np.concatenate((body, pred_a), axis=0)
+    pred_b = np.concatenate((body, pred_b), axis=0)
+    colors = np.concatenate((np.ones_like(body, dtype=np.int32)*100, np.ones_like(mesh_gt, dtype=np.int32)*230), axis=0)
+    # colors = np.ones_like(mesh_gt, dtype=np.int32)*230
+
+    fig.add_trace(go.Mesh3d(
+            x=mesh_gt[:,0],
+            y=mesh_gt[:,2],
+            z=mesh_gt[:,1],
+            i = faces[:,0],
+            j = faces[:,1],
+            k = faces[:,2],
+            text="Ground Truth",
+            vertexcolor=colors,
+        ),
+        row=1, col=1)
+
+    fig.add_trace(go.Mesh3d(
+            x=pred_a[:,0],
+            y=pred_a[:,2],
+            z=pred_a[:,1],
+            i = faces[:,0],
+            j = faces[:,1],
+            k = faces[:,2],
+            text="Pred A",
+            vertexcolor=colors,
+        ),
+        row=1, col=2)
+
+    fig.add_trace(go.Mesh3d(
+            x=pred_b[:,0],
+            y=pred_b[:,2],
+            z=pred_b[:,1],
+            i = faces[:,0],
+            j = faces[:,1],
+            k = faces[:,2],
+            text="Pred B",
+            vertexcolor=colors,
+        ),
+        row=1, col=3)
+
+
+
+    fig.update_layout(scene_aspectmode='manual', scene_aspectratio=dict(x=1, y=1, z=1), 
+                    scene2_aspectmode='manual', scene2_aspectratio=dict(x=1, y=1, z=1), 
+                    scene3_aspectmode='manual', scene3_aspectratio=dict(x=1, y=1, z=1), 
+                    )
+    helper = dict(xaxis = dict(nticks=4, range=[-1,1],),
+                yaxis = dict(nticks=4, range=[-1,1],),
+                zaxis = dict(nticks=4, range=[-1,1],),)
+    fig.update_layout(scene = helper,
+                        scene2 = helper,
+                        scene3 = helper,
+                        )
+
+    return fig
+
+
+
+if __name__ == '__main__':
+    run = wandb.init(project='plots', name=config_a['name'], notes=config_a['notes'], dir='/wandb')
+
+    parser = argparse.ArgumentParser(description='')
+    parser.add_argument('--name', type=str)
+    args = vars(parser.parse_args())
+
+    skirt_modifier = '_skirt' if config_a['is_skirt'] else ''
+
+    folder = config_a['test_dir']
+    # smpl = SMPL(use_one_hot=False)
+
+    F = np.load("custom_utils/smpl_faces" + skirt_modifier + ".npy").astype(np.int32)
+
+    iteration = ''
+
+    # load_model(model, 'checkpoints/' + 'abl_all_120k')
+
+
+    factory = DatasetFactory()
+    dataset_a = factory.get_dataset(folder=config_a['test_dir'], 
+                                       batch_size=config_a['batch_size'], 
+                                       signature=config_a['signature_dynamic'], 
+                                       resolutions=[config_a['resolution']], 
+                                       in_memory=False, 
+                                       do_repeat=False,
+                                       use_one_hot=config_a['use_one_hot'],
+                                       use_no_orient=config_a['remove_global_orientation'],
+                                       shuffle=False,
+                                       testing=True,
+                                    #    frames_file='frames_abl.txt',
+                                       )
+    static_dataset_a = StaticDataset(folder=config_a['test_dir'], 
+                                        signature=config_a['signature_static'], 
+                                        resolutions=[config_a['resolution']], 
+                                        in_memory=True)
+
+    iterator_a = iter(dataset_a)
+    
+
+    faces_folder = '/home/workspace/data/results/support/faces' + ('_skirt' if config_a['is_skirt'] else '')
+    ids = os.listdir(faces_folder)
+    ids = [id.split('.')[0] for id in ids]
+    faces = {id : np.load(faces_folder + '/' + id + '.npy') for id in ids }
+
+    results_folder = '/home/workspace/data/results/post'
+
+    count = 0
+    while True:
+        batch = iterator_a.get_next()
+
+        batch = static_dataset_a.expand_batch(batch_to_dict(batch, config_a['signature_dynamic']))
+
+        mask = batch['mesh_mask'].numpy().astype(bool)
+        gt = batch['mesh_posed'].numpy()
+        body = batch['mesh_body_posed'].numpy()
+
+        for i in range(mask.shape[0]):
+            id, frame_nr = batch['id'][i].numpy().decode("utf-8"), batch['frame_nr'][i].numpy().decode("utf-8")
+            pred_a = np.load(results_folder + '/al_all_post-/' + id + '_' + frame_nr + '.npy')
+            pred_b = np.load(results_folder + '/iccv_weights_test_6-/' + id + '_' + frame_nr + '.npy')
+            # pred_c = np.load(results_folder + '/iccv_weights_test_6-/' + id + '_' + frame_nr + '.npy')
+
+            fig = plot(gt[i], pred_a, pred_b, body[i], mask[i])
+
+            wandb.log({f'{id}_{frame_nr}': fig}, step = count)
+            count += 1
+
+            print(count)
+

configs/config_abl_all_post.py

@@ -7,7 +7,7 @@ config = dict()
 # DON'T FORGET TO LOAD FINAL UPSCALER UNLESS HAVE LEVEL 512
 # DON'T FORGET TO CHECK AVG GARMENT USED
 
-config['use_skinning'] =              True # and deshaping
+config['use_skinning'] =              False # and deshaping
 config['remove_global_orientation'] = True
 config['use_curvature'] =             True
 config['use_extra_losses'] =          True # edge, collision, 3d mesh
@@ -19,9 +19,9 @@ config['only_curvatures'] =           False
 ablation_count = [config['use_skinning'], config['remove_global_orientation'], config['use_curvature'], config['use_extra_losses'], config['use_offsets']].count(False)
 assert ablation_count == 1 or ablation_count == 0
 
-config['gpu'] = '4'
+config['gpu'] = '5'
-config['name'] = 'al_all_post'
+config['name'] = 'iccv_weights_test_8'
-config['notes'] = ''
+config['notes'] = 'smaller weight loss coeff, 1e-6, convolutional weight learning, changed convnext reverse'
 
 config['avg_garment_path'] = 'custom_data/train_no_shape_avg_uv.npy' #custom_data/train_no_shape_avg_uv.npy # train_no_shape_avg_uv  train_no_shape_one_hot_avg_uv
 if config['use_skinning']:
@@ -86,14 +86,15 @@ config['use_one_hot'] = False
 config['cvae'] = dict()
 config['cvae']['is_label'] = False
 config['cvae']['add_noise'] = True
-config['cvae']['noise_amount'] = 0.001
+config['cvae']['noise_amount'] = 0.00001
 
 config['cvae']['kernel_size'] = 7
 config['cvae']['encoder_depths'] = [3, 3, 3, 9, 3]
 config['cvae']['encoder_dims'] = [32, 64, 128, 256, 64]
-config['cvae']['decoder_depths'] = config['cvae']['encoder_depths']
+config['cvae']['decoder_depths'] = [3, 9, 3, 3, 3, 3]
-config['cvae']['decoder_dims'] = config['cvae']['encoder_dims']
+config['cvae']['decoder_dims'] = [3, 32, 128, 128, 256, 512]
-
+config['cvae']['decoder_weights_depths'] = [3, 9, 3, 3, 3]
+config['cvae']['decoder_weights_dims'] = [24, 32, 64, 128, 256]
 
 config['cond_ae_static'] = dict()
 config['cond_ae_static']['is_label'] = False

convnext/models/convnext_tf.py

@@ -2,7 +2,7 @@ from unittest import skip
 import numpy as np
 import tensorflow as tf
 from tensorflow.keras.layers import (Conv2D, Conv2DTranspose, Dense, DepthwiseConv2D,
-                                     GlobalAveragePooling2D, Layer,
+                                     GlobalAveragePooling2D, Layer, Activation, Lambda,
                                      LayerNormalization, Flatten, Reshape, UpSampling2D)
 
 def drop_path(inputs, drop_prob, is_training):
@@ -222,29 +222,31 @@ class ConvNeXtReverse(tf.keras.Model):
 
     def __init__(self, 
                  depths=[3, 3, 9, 3], dims=[96, 192, 384, 768], kernel_size=7,
-                 drop_path_rate=0., layer_scale_init_value=1e-6, head_init_scale=1.,
+                 drop_path_rate=0., layer_scale_init_value=1e-6, head_init_scale=1., learn_weights=False,
                  ):
         super().__init__()
-        self.upsample_layers = []  # stem and 3 intermediate downsampling conv layers
+        self.upsample_layers = [] 
+
         stem = tf.keras.Sequential([
-            Conv2DTranspose(dims[0], kernel_size=4, strides=4, padding='same'),
+            LayerNormalization(epsilon=1e-6),
-            # UpSampling2D((4, 4), interpolation='nearest'),
+            Conv2DTranspose(dims[0], kernel_size=2, strides=2, padding='same'),
-            # Conv2D(dims[0], kernel_size=4, strides=1, padding='same'),
+        ])
-            # Dense(dims[0]),
-            LayerNormalization(epsilon=1e-6)]
-        )
         self.upsample_layers.append(stem)
-        for i in range(len(depths) - 1):
+
+        for i in range(len(depths) - 2):
             downsample_layer = tf.keras.Sequential([
                 LayerNormalization(epsilon=1e-6),
-                Conv2DTranspose(dims[i+1], kernel_size=2, strides=2, padding='same')
+                Conv2DTranspose(dims[i+1], kernel_size=2, strides=2, padding='same'),
-                # UpSampling2D((2,2), interpolation='nearest'),
-                # Conv2D(dims[i+1], kernel_size=2, strides=1, padding='same'),
-                # Dense(dims[i+1]),
             ])
             self.upsample_layers.append(downsample_layer)
 
-        self.stages = []  # 4 feature resolution stages, each consisting of multiple residual blocks
+        downsample_layer = tf.keras.Sequential([
+            Conv2DTranspose(dims[i+2], kernel_size=2, strides=2, padding='same'),
+            LayerNormalization(epsilon=1e-6),
+        ])
+        self.upsample_layers.append(downsample_layer)
+
+        self.stages = []  
         dp_rates = [x for x in np.linspace(0, drop_path_rate, sum(depths))]
         cur = 0
         for i in range(len(depths)):
@@ -255,12 +257,19 @@ class ConvNeXtReverse(tf.keras.Model):
             self.stages.append(stage)
             cur += depths[i]
 
+        if not learn_weights:
             self.last = Dense(3, activation='tanh')
+            # self.last = Activation('tanh')
+        else:
+            self.last = Activation('softmax')
+            # self.last = Dense(24, activation='softmax')
+
 
     def call(self, x, training):
         for i in reversed(range(len(self.upsample_layers))):
             x = self.upsample_layers[i](x, training=training)
             x = self.stages[i](x, training=training)
+
         x = self.last(x, training=training)
         return x
 

custom_utils/custom_utils.py

@@ -2,6 +2,7 @@ import json
 import numpy as np
 import tensorflow as tf
 import scipy as sp
+import cmapy as cm
 import plotly.graph_objects as go
 from plotly.subplots import make_subplots
 
@@ -216,6 +217,35 @@ def plot_3d(mesh_gt, mesh_pred_unposed, faces, collision, collision_faces, colli
     return fig
 
 
+def plot_weights(weights, body, faces):
+    fig = go.Figure()
+    colors = np.array([cm.color('jet', 1 - w) for w in weights.numpy()])
+
+    # if flip_xy:
+    #     mesh_gt = flip_axis(mesh_gt)
+    #     mesh_pred_unposed = flip_axis(mesh_pred_unposed)
+    #     collision = flip_axis(collision, ind=14475)
+
+    fig.add_trace(go.Mesh3d(
+                x=body[:,0],
+                y=body[:,2],
+                z=body[:,1],
+                i = faces[:,0],
+                j = faces[:,1],
+                k = faces[:,2],
+                vertexcolor=colors
+            ))
+    
+
+    fig.update_layout(scene_aspectmode='manual', scene_aspectratio=dict(x=1, y=1, z=1),)
+    helper = dict(xaxis = dict(nticks=4, range=[-1,1],),
+                yaxis = dict(nticks=4, range=[-1,1],),
+                zaxis = dict(nticks=4, range=[-1,1],),)
+    fig.update_layout(scene = helper,)
+
+    return fig
+
+
 def plot_3d_collision(gt, collision_1, collision_2, collision_faces, collision_color, flip_xy=False):
     fig = make_subplots(rows=1, cols=3, start_cell="top-left", specs = [ [{ 'type': 'scene' } for _ in range(3)] ])
 

custom_utils/smpl.py

@@ -406,7 +406,8 @@ class SMPL:
         return T_out, J_out
 
 
-    def run_garment_with_weights(self, R, betas, v_cloth, weights, J=None):
+    def run_garment_with_weights(self, pose, betas, v_cloth, weights, J=None):
+        R = self.run_rodrigues(pose)
         
         if J is None:
             J = self.get_joints_template(betas)
@@ -430,13 +431,12 @@ class SMPL:
             J = tf.gather(J, tf.constant([0, 1, 2], tf.int32), axis=1)
             
             T = tf.tensordot(weights, results, axes=((1), (0)))
-            rest_shape_h = tf.concat((v_cloth, tf.ones((v_cloth.get_shape().as_list()[0], 1), dtype=tf.float32)), axis=1)
+            rest_shape_h = tf.concat((v_cloth, tf.ones((v_cloth.shape[0], 1), dtype=tf.float32)), axis=1)
             v = tf.matmul(T, tf.reshape(rest_shape_h, (-1, 4, 1)))
             v = tf.reshape(v, (-1, 4))[:, :3]
 
             return v, tf.squeeze(J)
 
-        # gender = tf.cast(tf.greater_equal(betas[:, 10] , 0), tf.int32)
         T_out, J_out = tf.map_fn(single_run, (R, J, v_cloth, weights), (tf.float32, tf.float32))
         
         J0 = tf.reshape(J_out[:,0,:], (-1, 1, 3))

cvae.py

@@ -38,7 +38,7 @@ else:
 from losses import collision_ratio, kl_loss, normal_loss #, concept_loss
 # from losses import collision_penalty as collision_penalty
 from losses import collision_loss as collision_penalty
-from custom_utils.custom_utils import batch_to_dict, load_model, plot_3d, UV2Mesh, get_summary_html, maskFacesBatch, _maskFaces, plot_3d_collision
+from custom_utils.custom_utils import batch_to_dict, load_model, plot_3d, UV2Mesh, UV2MeshLabel, get_summary_html, maskFacesBatch, _maskFaces, plot_3d_collision, plot_weights
 from dataset import DatasetFactory, StaticDataset
 from custom_utils.smpl import SMPL
 from models.cvae_model import CVAENew as CVAE
@@ -62,6 +62,7 @@ def init_wandb():
     run.log_artifact(artifact)
 
 uv2mesh = UV2Mesh(res=256, is_skirt=config['is_skirt'], is_adaptive=False, is_smpl_orig=config['is_cape'])
+uv2mesh_label = UV2MeshLabel(res=256, is_skirt=config['is_skirt'], is_smpl_orig=config['is_cape'])
 
 
 def kl_loss(mean, logvar):
@@ -71,9 +72,15 @@ def kl_loss(mean, logvar):
     return kl_loss
 
 
+def blend_weights_reg(pred_w, gender=None):
+    orig = smpl.weights
+    l1 = tf.reduce_mean(tf.abs(pred_w - orig[None, 0, :, :]))
+    return l1
+
+
 # 3d loss on posed !!!!!!!!!!!!
 
-# @tf.function
+@tf.function
 def train_step(batch):
 
     with tf.GradientTape() as tape:
@@ -82,9 +89,10 @@ def train_step(batch):
         else:
             mesh_orig, uv_orig = batch['mesh_posed'], batch['uv_posed']
 
-        pred, mean, logvar = model([uv_orig, batch['uv_static'], batch['uv_body_posed'], batch['uv_curvature'], batch['uv_mask']], training=True)
+        pred, mean, logvar, weights = model([uv_orig, batch['uv_static'], batch['uv_body_posed'], batch['uv_curvature'], batch['uv_mask']], training=True)
 
         mesh_pred = uv2mesh(pred)
+        mesh_weigths = uv2mesh_label(weights)
 
         if config['use_skinning']:
             shape = batch['pose_shape'][:, 72:]
@@ -94,25 +102,31 @@ def train_step(batch):
             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
+            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]
+            mesh_pred_posed = smpl.run_garment_with_weights(batch['pose_shape'][:, :72], batch['pose_shape'][:, 72:], mesh_pred_shaped, mesh_weigths)
 
         # is the output already masked?
 
         e_loss = tf.reduce_mean(edge_loss(mesh_pred, mesh_orig, batch['mesh_mask']))
         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]))
+        # l2_loss = tf.reduce_mean(1e3 * tf.reduce_mean(tf.reduce_sum(((uv_orig - pred) ** 2), axis=-1), axis=[1, 2]))
         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'], is_smpl_orig=config['is_cape'])
         # conc_loss = concept_loss(pred, batch['uv_unposed'])
         kl = kl_loss(mean, logvar)
+        weights_l = blend_weights_reg(mesh_weigths)
 
-        reconstr_loss = l2_loss
+        reconstr_loss = l2_mesh_loss + 1e-6 * weights_l
         if config['use_extra_losses']:
             if 'block_collision_loss' in config.keys() and config['block_collision_loss'] == True:
-                reconstr_loss = reconstr_loss + l2_mesh_loss + (e_loss) * config['reg_loss_coeff']
+                reconstr_loss = reconstr_loss + (e_loss) * config['reg_loss_coeff']
             else:
-                reconstr_loss = reconstr_loss + l2_mesh_loss + (e_loss + c_loss) * config['reg_loss_coeff']
+                reconstr_loss = reconstr_loss + (e_loss + c_loss) * config['reg_loss_coeff']
         if config['use_norm_loss'] and step > 2000:
             reconstr_loss = reconstr_loss + n_loss * config['reg_loss_coeff']
 
@@ -122,14 +136,14 @@ def train_step(batch):
         logvar = tf.reduce_mean(logvar)
 
     # 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)))
-    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'])
+    metric = tf.reduce_sum(tf.math.sqrt(tf.reduce_sum(batch['mesh_mask'][:,:,tf.newaxis] * (mesh_orig - mesh_pred_posed) ** 2, axis=-1))) / tf.reduce_sum(batch['mesh_mask'])
     gradients = tape.gradient(loss, model.trainable_variables)
     optimizer.apply_gradients(zip(gradients, model.trainable_variables))
     return { 'train/loss': loss, 'train/edge_loss': e_loss, 'train/collision_loss': c_loss, 'train/euclid_error': metric,
              'train/reconstr_loss': reconstr_loss, 'train/collision_ratio': c_ratio,
              'train/mean': mean, 'train/logvar': logvar, 'train/kl_loss': kl, 'train/var': np.e ** logvar,
-             'train/l2_loss': l2_loss, 'train/l2_mesh_loss': l2_mesh_loss, 'train/normal_loss': n_loss,
+             'train/l2_mesh_loss': l2_mesh_loss, 'train/normal_loss': n_loss, 'train/weights_loss': weights_l,
-             # 'train/ms_ssim': ms_ssim, 
+             # 'train/ms_ssim': ms_ssim, 'train/l2_loss': l2_loss, 
             }
 
 
@@ -140,9 +154,10 @@ def valid_step(batch):
     else:
         mesh_orig, uv_orig = batch['mesh_posed'], batch['uv_posed']
 
-    pred, mean, logvar = model([uv_orig, batch['uv_static'], batch['uv_body_posed'], batch['uv_curvature'], batch['uv_mask']], training=False)
+    pred, mean, logvar, weights = model([uv_orig, batch['uv_static'], batch['uv_body_posed'], batch['uv_curvature'], batch['uv_mask']], training=False)
 
     mesh_pred = uv2mesh(pred)
+    mesh_weigths = uv2mesh_label(weights)
 
     if config['use_skinning']:
         shape = batch['pose_shape'][:, 72:]
@@ -152,22 +167,28 @@ def valid_step(batch):
         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
+        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]
+        mesh_pred_posed = smpl.run_garment_with_weights(batch['pose_shape'][:, :72], batch['pose_shape'][:, 72:], mesh_pred_shaped, mesh_weigths)
 
     # is the output already masked?
 
     e_loss = tf.reduce_mean(edge_loss(mesh_pred, mesh_orig, batch['mesh_mask']))
     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]))
+    # l2_loss = tf.reduce_mean(1e3 * tf.reduce_mean(tf.reduce_sum(((uv_orig - pred) ** 2), axis=-1), axis=[1, 2]))
     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'], is_smpl_orig=config['is_cape'])
     # conc_loss = concept_loss(pred, batch['uv_unposed'])
     kl = kl_loss(mean, logvar)
+    weights_l = blend_weights_reg(mesh_weigths)
 
-    reconstr_loss = l2_loss
+    reconstr_loss = l2_mesh_loss + 1e-6 * weights_l
     if config['use_extra_losses']:
-        reconstr_loss = reconstr_loss + l2_mesh_loss + (e_loss + c_loss) * config['reg_loss_coeff']
+        reconstr_loss = reconstr_loss + (e_loss + c_loss) * config['reg_loss_coeff']
     if config['use_norm_loss']:
         reconstr_loss = reconstr_loss + n_loss * config['reg_loss_coeff']
 
@@ -177,12 +198,12 @@ def valid_step(batch):
     logvar = tf.reduce_mean(logvar)
 
     # 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)))
-    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'])
+    metric = tf.reduce_sum(tf.math.sqrt(tf.reduce_sum(batch['mesh_mask'][:,:,tf.newaxis] * (mesh_orig - mesh_pred_posed) ** 2, axis=-1))) / tf.reduce_sum(batch['mesh_mask'])
     # c_ratio = tf.reduce_mean(collision_ratio(mesh_pred_posed, batch['mesh_body_posed'], batch['mesh_mask']))
 
     return { 'valid/loss': loss, 'valid/edge_loss': e_loss, 'valid/collision_loss': c_loss, 'valid/euclid_error': metric, 
              'valid/reconstr_loss': reconstr_loss, 'valid/l2_mesh_loss': l2_mesh_loss, 'valid/collision_ratio': c_ratio,
-             'valid/mean': mean, 'valid/logvar': logvar, 'valid/kl_loss': kl, 'valid/var': np.e ** logvar,
+             'valid/mean': mean, 'valid/logvar': logvar, 'valid/kl_loss': kl, 'valid/var': np.e ** logvar, 'valid/weights_loss': weights_l,
               'valid/normal_loss': n_loss,# 'valid/ms_ssim': ms_ssim, 
               }
 
@@ -194,7 +215,7 @@ def test_step(batch):
     else:
         mesh_orig, uv_orig = batch['mesh_posed'], batch['uv_posed']
     
-    pred, mean, logvar = model([uv_orig, batch['uv_static'], batch['uv_body_posed'], batch['uv_curvature'], batch['uv_mask']], training=False)
+    pred, mean, logvar, weights = model([uv_orig, batch['uv_static'], batch['uv_body_posed'], batch['uv_curvature'], batch['uv_mask']], training=False)
 
     mesh_pred = uv2mesh(pred)
 
@@ -208,7 +229,7 @@ def test_step(batch):
         mesh_orig_shaped = mesh_orig + shape_offsets * batch['mesh_mask'][:,:,tf.newaxis]
         mesh_orig_posed = smpl.run_garment(batch['pose_shape'][:, :72], batch['pose_shape'][:, 72:], mesh_orig_shaped)
     else:
-        mesh_pred_posed = mesh_pred
+        mesh_pred_posed = smpl.run_garment_with_weights(batch['pose_shape'][:, :72], batch['pose_shape'][:, 72:], mesh_pred, weights)
         mesh_orig_posed = mesh_orig
     
     mesh_pred_posed_fixed = fix_collision(mesh_pred_posed, batch['mesh_body_posed'], batch['mesh_mask'])
@@ -250,9 +271,10 @@ def plot_everything(batch: dict, type: str):
     else:
         mesh_orig, uv_orig = batch['mesh_posed'], batch['uv_posed']
 
-    pred, mean, logvar = model([uv_orig, batch['uv_static'], batch['uv_body_posed'], batch['uv_curvature'], batch['uv_mask']], training=False)
+    pred, mean, logvar, weights = model([uv_orig, batch['uv_static'], batch['uv_body_posed'], batch['uv_curvature'], batch['uv_mask']], training=False)
     
     mesh_pred = uv2mesh(pred).numpy()
+    mesh_weigths = uv2mesh_label(weights)
     
     orig_img = uv_orig
     uv_static_img = batch['uv_static']
@@ -277,7 +299,12 @@ def plot_everything(batch: dict, type: str):
             mesh_pred[i, mask[i], :] += shape_offsets.numpy()[i, mask[i], :]
         mesh_posed = smpl.run_garment(batch['pose_shape'][:, :72], batch['pose_shape'][:, 72:], mesh_pred).numpy()
     else:
-        mesh_posed = mesh_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]
+        mesh_posed = smpl.run_garment_with_weights(batch['pose_shape'][:, :72], batch['pose_shape'][:, 72:], mesh_pred_shaped, mesh_weigths).numpy()
         gt = [mesh_orig[i].numpy()[mask[i], :] for i in range(pred.shape[0])]
         garment_unposed, faces = maskFacesBatch(mesh_pred, F, mask)
 
@@ -296,13 +323,16 @@ def plot_everything(batch: dict, type: str):
     collision_fixed = [np.concatenate((body[i], garment_posed_fixed[i]), axis=0) for i in range(len(garment_posed))]
     collision_gt = [np.concatenate((body[i], gt_posed[i]), axis=0) for i in range(len(garment_posed))]
 
-    figs = { type + "/garment_" + str(i): plot_3d(gt[i], garment_unposed[i], faces[i], collision[i], collision_faces[i], collision_color[i], flip_xy=(not config['use_skinning'] and config['remove_global_orientation'])) for i in range(4) }
+    # figs = { type + "/garment_" + str(i): plot_3d(gt[i], garment_unposed[i], faces[i], garment_posed[i], faces[i], collision_color[i], flip_xy=(not config['use_skinning'] and config['remove_global_orientation'])) for i in range(4) }
-    figs_coll = { type + "/garment_" + str(i) + '_collision': plot_3d_collision(collision_gt[i], collision[i], collision_fixed[i], collision_faces[i], collision_color[i], flip_xy=(not config['use_skinning'] and config['remove_global_orientation'])) for i in range(4) }
+    figs = { type + "/garment_" + str(i): plot_3d(gt[i], garment_unposed[i], faces[i], collision[i], collision_faces[i], collision_color[i],) for i in range(4) }
+    figs_coll = { type + "/garment_" + str(i) + '_collision': plot_3d_collision(collision_gt[i], collision[i], collision_fixed[i], collision_faces[i], collision_color[i],) for i in range(4) }
+    figs_weights = { type + '/weights_' + str(i): plot_weights(mesh_weigths[0,:,i], body[0], F) for i in range(24) }
+    # figs_weights = { type + '/weights_' + str(i): plot_weights(smpl.weights[0,:,i], body[0], F) for i in range(24) }
 
     mean_hist = wandb.Histogram(sequence=mean.numpy().mean(axis=0))
     logvar_hist = wandb.Histogram(sequence=logvar.numpy().mean(axis=0))
 
-    return { **figs, **figs_coll, **{ type + '/images': imgs, type + '/mean_hist': mean_hist, type + '/logvar_hist': logvar_hist } }
+    return { **figs, **figs_coll, **figs_weights, **{ type + '/images': imgs, type + '/mean_hist': mean_hist, type + '/logvar_hist': logvar_hist } }
 
 
 def sample(folder): 
@@ -476,7 +506,7 @@ def test():
 
 
 if __name__ == '__main__':
-    # init_wandb()
+    init_wandb()
     parser = argparse.ArgumentParser(description='')
     parser.add_argument('--name', type=str)
     args = vars(parser.parse_args())

evaluate_ae.py

@@ -39,7 +39,7 @@ else:
 from losses import collision_ratio, kl_loss #, concept_loss
 # from losses import collision_penalty as collision_penalty
 from losses import collision_loss as collision_penalty
-from custom_utils.custom_utils import batch_to_dict, load_model, plot_3d, UV2Mesh, get_summary_html, maskFacesBatch, _maskFaces, plot_3d_collision
+from custom_utils.custom_utils import batch_to_dict, load_model, plot_3d, UV2Mesh, UV2MeshLabel, get_summary_html, maskFacesBatch, _maskFaces, plot_3d_collision
 from dataset import DatasetFactory, StaticDataset
 from custom_utils.smpl import SMPL
 from models.cvae_model import CVAENew as CVAE
@@ -48,6 +48,7 @@ from custom_utils.supersmpl_io import writeOBJ
 
 
 uv2mesh = UV2Mesh(res=256, is_skirt=config['is_skirt'], is_adaptive=False, is_smpl_orig=config['is_cape'])
+uv2mesh_label = UV2MeshLabel(res=256, is_skirt=config['is_skirt'], is_smpl_orig=config['is_cape'])
 
 @tf.function
 def step(batch):
@@ -56,9 +57,10 @@ def step(batch):
     else:
         uv_orig =  batch['uv_posed']
     
-    pred, _, _ = model([uv_orig, batch['uv_static'], batch['uv_body_posed'], batch['uv_curvature'], batch['uv_mask']], training=False)
+    pred, _, _, weights = model([uv_orig, batch['uv_static'], batch['uv_body_posed'], batch['uv_curvature'], batch['uv_mask']], training=False)
 
     mesh_pred = uv2mesh(pred)
+    mesh_weigths = uv2mesh_label(weights)
 
     if config['use_skinning']:
         shape = batch['pose_shape'][:, 72:]
@@ -68,7 +70,13 @@ def step(batch):
         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
+        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]
+        mesh_pred_posed = smpl.run_garment_with_weights(batch['pose_shape'][:, :72], batch['pose_shape'][:, 72:], mesh_pred_shaped, mesh_weigths)
+
         
     # mesh_pred_posed_fixed = fix_collision(mesh_pred_posed, batch['mesh_body_posed'], batch['mesh_mask'])
 

models/base.py

@@ -171,13 +171,13 @@ class AEEncoderNew(tf.keras.Model):
 
 class DecoderNew(tf.keras.Model):
     
-    def __init__(self, depths, dims, kernel_size=7, is_label=False):
+    def __init__(self, depths, dims, kernel_size=7, is_label=False, learn_weights=False):
         super(DecoderNew, self).__init__()
 
         self.is_label = is_label
 
         if not is_label:
-            self.decoder = ConvNeXtReverse(depths=depths, dims=dims, kernel_size=kernel_size)
+            self.decoder = ConvNeXtReverse(depths=depths, dims=dims, kernel_size=kernel_size, learn_weights=learn_weights)
         else:
             self.decoder = ConvNeXtReverseLabels(depths=depths, dims=dims, nr_of_classes=7)
             

models/cvae_model.py

@@ -181,6 +181,7 @@ class CVAENew(tf.keras.Model):
 
         self.encoder = VariationalEncoderNew(config['cvae']['encoder_depths'], config['cvae']['encoder_dims'], kernel_size=config['cvae']['kernel_size'])
         self.decoder = DecoderNew(config['cvae']['decoder_depths'], config['cvae']['decoder_dims'], config['cvae']['kernel_size'])
+        self.decoder_weights = DecoderNew(config['cvae']['decoder_weights_depths'], config['cvae']['decoder_weights_dims'], config['cvae']['kernel_size'], learn_weights=True)
 
         self.reshape = tf.keras.layers.Reshape((8, 4, 64))
 
@@ -193,12 +194,12 @@ class CVAENew(tf.keras.Model):
         self.noise_amount = config['cvae']['noise_amount']
         self.use_curvature = config['use_curvature']
         self.only_curvatures = config['only_curvatures'] if 'only_curvatures' in config.keys() and config['only_curvatures'] == True else None
-        self.use_offsets_over_body = not config['use_skinning']
+        self.use_offsets_over_body = False
 
         if self.use_offsets_over_body:
             assert self.avg_garment is None, 'When we are not using skinning, we can\'t use avg_garment!'
-        if self.avg_garment is not None:
+        # if self.avg_garment is not None:
-            assert not self.use_offsets_over_body, 'When avg_garment is given, we must use skinning!'
+        #     assert not self.use_offsets_over_body, 'When avg_garment is given, we must use skinning!'
 
     # @tf.function
     def sample(self, uv_template, uv_body_posed, uv_curv, uv_mask, noise=None, add_noise=False):
@@ -214,6 +215,7 @@ class CVAENew(tf.keras.Model):
         z = tf.concat([eps, z_cond], axis=-1)
 
         x = self.decoder(z, training=False)
+        weights = self.decoder_weights(z, training=False)
         
         if self.avg_garment is not None:
             x = x + self.avg_garment[tf.newaxis,:,:,:]
@@ -223,7 +225,7 @@ class CVAENew(tf.keras.Model):
 
         x = x * uv_mask[:,:,:,tf.newaxis]
 
-        return x
+        return x, weights
 
 
     def reparameterize(self, mean, logvar, training, fix_noise=False):
@@ -272,6 +274,7 @@ class CVAENew(tf.keras.Model):
 
         z = tf.concat([z, z_cond], axis=-1)
         x = self.decoder(z, training=training)
+        weights = self.decoder_weights(z, training=training)
 
         if self.avg_garment is not None:
             x = x + self.avg_garment[tf.newaxis,:,:,:]
@@ -281,5 +284,5 @@ class CVAENew(tf.keras.Model):
 
         x = x * uv_mask[:,:,:,tf.newaxis]
 
-        return x, mean, logvar
+        return x, mean, logvar, weights