diff --git a/mainscripts/Trainer.py b/mainscripts/Trainer.py index a2625d0..fc8283b 100644 --- a/mainscripts/Trainer.py +++ b/mainscripts/Trainer.py @@ -40,30 +40,33 @@ def trainerThread (s2c, c2s, args, device_args): debug=debug, device_args=device_args) - is_reached_goal = model.is_reached_epoch_goal() + is_reached_goal = model.is_reached_iter_goal() is_upd_save_time_after_train = False + loss_string = "" def model_save(): if not debug and not is_reached_goal: + io.log_info ("Saving....", end='\r') model.save() + io.log_info(loss_string) is_upd_save_time_after_train = True def send_preview(): if not debug: previews = model.get_previews() - c2s.put ( {'op':'show', 'previews': previews, 'epoch':model.get_epoch(), 'loss_history': model.get_loss_history().copy() } ) + c2s.put ( {'op':'show', 'previews': previews, 'iter':model.get_iter(), 'loss_history': model.get_loss_history().copy() } ) else: - previews = [( 'debug, press update for new', model.debug_one_epoch())] + previews = [( 'debug, press update for new', model.debug_one_iter())] c2s.put ( {'op':'show', 'previews': previews} ) if model.is_first_run(): model_save() - if model.get_target_epoch() != 0: + if model.get_target_iter() != 0: if is_reached_goal: - io.log_info('Model already trained to target epoch. You can use preview.') + io.log_info('Model already trained to target iteration. You can use preview.') else: - io.log_info('Starting. Target epoch: %d. Press "Enter" to stop training and save model.' % ( model.get_target_epoch() ) ) + io.log_info('Starting. Target iteration: %d. Press "Enter" to stop training and save model.' % ( model.get_target_iter() ) ) else: io.log_info('Starting. Press "Enter" to stop training and save model.') @@ -72,14 +75,14 @@ def trainerThread (s2c, c2s, args, device_args): for i in itertools.count(0,1): if not debug: if not is_reached_goal: - loss_string = model.train_one_epoch() + loss_string = model.train_one_iter() if is_upd_save_time_after_train: #save resets plaidML programs, so upd last_save_time only after plaidML rebuild them last_save_time = time.time() io.log_info (loss_string, end='\r') - if model.get_target_epoch() != 0 and model.is_reached_epoch_goal(): - io.log_info ('Reached target epoch.') + if model.get_target_iter() != 0 and model.is_reached_iter_goal(): + io.log_info ('Reached target iteration.') model_save() is_reached_goal = True io.log_info ('You can use preview now.') @@ -91,7 +94,7 @@ def trainerThread (s2c, c2s, args, device_args): if i==0: if is_reached_goal: - model.pass_one_epoch() + model.pass_one_iter() send_preview() if debug: @@ -104,7 +107,7 @@ def trainerThread (s2c, c2s, args, device_args): model_save() elif op == 'preview': if is_reached_goal: - model.pass_one_epoch() + model.pass_one_iter() send_preview() elif op == 'close': model_save() @@ -156,8 +159,8 @@ def main(args, device_args): update_preview = False is_showing = False is_waiting_preview = False - show_last_history_epochs_count = 0 - epoch = 0 + show_last_history_iters_count = 0 + iter = 0 while True: if not c2s.empty(): input = c2s.get() @@ -166,7 +169,7 @@ def main(args, device_args): is_waiting_preview = False loss_history = input['loss_history'] if 'loss_history' in input.keys() else None previews = input['previews'] if 'previews' in input.keys() else None - epoch = input['epoch'] if 'epoch' in input.keys() else 0 + iter = input['iter'] if 'iter' in input.keys() else 0 if previews is not None: max_w = 0 max_h = 0 @@ -217,12 +220,12 @@ def main(args, device_args): final = head if loss_history is not None: - if show_last_history_epochs_count == 0: + if show_last_history_iters_count == 0: loss_history_to_show = loss_history else: - loss_history_to_show = loss_history[-show_last_history_epochs_count:] + loss_history_to_show = loss_history[-show_last_history_iters_count:] - lh_img = models.ModelBase.get_loss_history_preview(loss_history_to_show, epoch, w, c) + lh_img = models.ModelBase.get_loss_history_preview(loss_history_to_show, iter, w, c) final = np.concatenate ( [final, lh_img], axis=0 ) final = np.concatenate ( [final, selected_preview_rgb], axis=0 ) @@ -243,16 +246,16 @@ def main(args, device_args): is_waiting_preview = True s2c.put ( {'op': 'preview'} ) elif key == ord('l'): - if show_last_history_epochs_count == 0: - show_last_history_epochs_count = 5000 - elif show_last_history_epochs_count == 5000: - show_last_history_epochs_count = 10000 - elif show_last_history_epochs_count == 10000: - show_last_history_epochs_count = 50000 - elif show_last_history_epochs_count == 50000: - show_last_history_epochs_count = 100000 - elif show_last_history_epochs_count == 100000: - show_last_history_epochs_count = 0 + if show_last_history_iters_count == 0: + show_last_history_iters_count = 5000 + elif show_last_history_iters_count == 5000: + show_last_history_iters_count = 10000 + elif show_last_history_iters_count == 10000: + show_last_history_iters_count = 50000 + elif show_last_history_iters_count == 50000: + show_last_history_iters_count = 100000 + elif show_last_history_iters_count == 100000: + show_last_history_iters_count = 0 update_preview = True elif key == ord(' '): selected_preview = (selected_preview + 1) % len(previews) diff --git a/models/ModelBase.py b/models/ModelBase.py index b1bc47c..ac3beee 100644 --- a/models/ModelBase.py +++ b/models/ModelBase.py @@ -51,53 +51,57 @@ class ModelBase(object): self.debug = debug self.is_training_mode = (training_data_src_path is not None and training_data_dst_path is not None) - self.epoch = 0 + self.iter = 0 self.options = {} self.loss_history = [] self.sample_for_preview = None if self.model_data_path.exists(): model_data = pickle.loads ( self.model_data_path.read_bytes() ) - self.epoch = model_data['epoch'] - if self.epoch != 0: + self.iter = max( model_data.get('iter',0), model_data.get('epoch',0) ) + if 'epoch' in self.options: + self.options.pop('epoch') + if self.iter != 0: self.options = model_data['options'] self.loss_history = model_data['loss_history'] if 'loss_history' in model_data.keys() else [] self.sample_for_preview = model_data['sample_for_preview'] if 'sample_for_preview' in model_data.keys() else None - ask_override = self.is_training_mode and self.epoch != 0 and io.input_in_time ("Press enter in 2 seconds to override model settings.", 2) + ask_override = self.is_training_mode and self.iter != 0 and io.input_in_time ("Press enter in 2 seconds to override model settings.", 2) yn_str = {True:'y',False:'n'} - if self.epoch == 0: + if self.iter == 0: io.log_info ("\nModel first run. Enter model options as default for each run.") - if self.epoch == 0 or ask_override: - default_write_preview_history = False if self.epoch == 0 else self.options.get('write_preview_history',False) + if self.iter == 0 or ask_override: + default_write_preview_history = False if self.iter == 0 else self.options.get('write_preview_history',False) self.options['write_preview_history'] = io.input_bool("Write preview history? (y/n ?:help skip:%s) : " % (yn_str[default_write_preview_history]) , default_write_preview_history, help_message="Preview history will be writed to _history folder.") else: self.options['write_preview_history'] = self.options.get('write_preview_history', False) - if self.epoch == 0 or ask_override: - self.options['target_epoch'] = max(0, io.input_int("Target epoch (skip:unlimited/default) : ", 0)) + if self.iter == 0 or ask_override: + self.options['target_iter'] = max(0, io.input_int("Target iteration (skip:unlimited/default) : ", 0)) else: - self.options['target_epoch'] = self.options.get('target_epoch', 0) + self.options['target_iter'] = max(model_data.get('target_iter',0), self.options.get('target_epoch',0)) + if 'target_epoch' in self.options: + self.options.pop('target_epoch') - if self.epoch == 0 or ask_override: - default_batch_size = 0 if self.epoch == 0 else self.options.get('batch_size',0) + if self.iter == 0 or ask_override: + default_batch_size = 0 if self.iter == 0 else self.options.get('batch_size',0) self.options['batch_size'] = max(0, io.input_int("Batch_size (?:help skip:0/default) : ", default_batch_size, help_message="Larger batch size is always better for NN's generalization, but it can cause Out of Memory error. Tune this value for your videocard manually.")) else: self.options['batch_size'] = self.options.get('batch_size', 0) - if self.epoch == 0: + if self.iter == 0: self.options['sort_by_yaw'] = io.input_bool("Feed faces to network sorted by yaw? (y/n ?:help skip:n) : ", False, help_message="NN will not learn src face directions that don't match dst face directions." ) else: self.options['sort_by_yaw'] = self.options.get('sort_by_yaw', False) - if self.epoch == 0: + if self.iter == 0: self.options['random_flip'] = io.input_bool("Flip faces randomly? (y/n ?:help skip:y) : ", True, help_message="Predicted face will look more naturally without this option, but src faceset should cover all face directions as dst faceset.") else: self.options['random_flip'] = self.options.get('random_flip', True) - if self.epoch == 0: + if self.iter == 0: self.options['src_scale_mod'] = np.clip( io.input_int("Src face scale modifier % ( -30...30, ?:help skip:0) : ", 0, help_message="If src face shape is wider than dst, try to decrease this value to get a better result."), -30, 30) else: self.options['src_scale_mod'] = self.options.get('src_scale_mod', 0) @@ -106,9 +110,9 @@ class ModelBase(object): if not self.options['write_preview_history']: self.options.pop('write_preview_history') - self.target_epoch = self.options['target_epoch'] - if self.options['target_epoch'] == 0: - self.options.pop('target_epoch') + self.target_iter = self.options['target_iter'] + if self.options['target_iter'] == 0: + self.options.pop('target_iter') self.batch_size = self.options['batch_size'] self.sort_by_yaw = self.options['sort_by_yaw'] @@ -118,7 +122,7 @@ class ModelBase(object): if self.src_scale_mod == 0: self.options.pop('src_scale_mod') - self.onInitializeOptions(self.epoch == 0, ask_override) + self.onInitializeOptions(self.iter == 0, ask_override) nnlib.import_all ( nnlib.DeviceConfig(allow_growth=False, **self.device_args) ) self.device_config = nnlib.active_DeviceConfig @@ -142,7 +146,7 @@ class ModelBase(object): if not self.preview_history_path.exists(): self.preview_history_path.mkdir(exist_ok=True) else: - if self.epoch == 0: + if self.iter == 0: for filename in Path_utils.get_image_paths(self.preview_history_path): Path(filename).unlink() @@ -153,7 +157,7 @@ class ModelBase(object): if not isinstance(generator, SampleGeneratorBase): raise ValueError('training data generator is not subclass of SampleGeneratorBase') - if (self.sample_for_preview is None) or (self.epoch == 0): + if (self.sample_for_preview is None) or (self.iter == 0): self.sample_for_preview = self.generate_next_sample() model_summary_text = [] @@ -161,7 +165,7 @@ class ModelBase(object): model_summary_text += ["===== Model summary ====="] model_summary_text += ["== Model name: " + self.get_model_name()] model_summary_text += ["=="] - model_summary_text += ["== Current epoch: " + str(self.epoch)] + model_summary_text += ["== Current iteration: " + str(self.iter)] model_summary_text += ["=="] model_summary_text += ["== Model options:"] for key in self.options.keys(): @@ -210,7 +214,7 @@ class ModelBase(object): pass #overridable - def onTrainOneEpoch(self, sample, generator_list): + def onTrainOneIter(self, sample, generator_list): #train your keras models here #return array of losses @@ -231,11 +235,11 @@ class ModelBase(object): raise NotImplementeError #return existing or your own converter which derived from base - def get_target_epoch(self): - return self.target_epoch + def get_target_iter(self): + return self.target_iter - def is_reached_epoch_goal(self): - return self.target_epoch != 0 and self.epoch >= self.target_epoch + def is_reached_iter_goal(self): + return self.target_iter != 0 and self.iter >= self.target_iter #multi gpu in keras actually is fake and doesn't work for training https://github.com/keras-team/keras/issues/11976 #def to_multi_gpu_model_if_possible (self, models_list): @@ -263,13 +267,13 @@ class ModelBase(object): return self.onGetPreview (self.sample_for_preview)[0][1] #first preview, and bgr def save(self): - io.log_info ("Saving...") + io.log_info ("Saving....", end='\r') Path( self.get_strpath_storage_for_file('summary.txt') ).write_text(self.model_summary_text) self.onSave() model_data = { - 'epoch': self.epoch, + 'iter': self.iter, 'options': self.options, 'loss_history': self.loss_history, 'sample_for_preview' : self.sample_for_preview @@ -336,7 +340,7 @@ class ModelBase(object): source_filename.rename ( str(target_filename) ) - def debug_one_epoch(self): + def debug_one_iter(self): images = [] for generator in self.generator_list: for i,batch in enumerate(next(generator)): @@ -348,42 +352,42 @@ class ModelBase(object): def generate_next_sample(self): return [next(generator) for generator in self.generator_list] - def train_one_epoch(self): + def train_one_iter(self): sample = self.generate_next_sample() - epoch_time = time.time() - losses = self.onTrainOneEpoch(sample, self.generator_list) - epoch_time = time.time() - epoch_time + iter_time = time.time() + losses = self.onTrainOneIter(sample, self.generator_list) + iter_time = time.time() - iter_time self.last_sample = sample self.loss_history.append ( [float(loss[1]) for loss in losses] ) if self.write_preview_history: - if self.epoch % 10 == 0: + if self.iter % 10 == 0: preview = self.get_static_preview() - preview_lh = ModelBase.get_loss_history_preview(self.loss_history, self.epoch, preview.shape[1], preview.shape[2]) + preview_lh = ModelBase.get_loss_history_preview(self.loss_history, self.iter, preview.shape[1], preview.shape[2]) img = (np.concatenate ( [preview_lh, preview], axis=0 ) * 255).astype(np.uint8) - cv2_imwrite ( str (self.preview_history_path / ('%.6d.jpg' %( self.epoch) )), img ) + cv2_imwrite ( str (self.preview_history_path / ('%.6d.jpg' %( self.iter) )), img ) - self.epoch += 1 + self.iter += 1 - if epoch_time >= 10: - #............."Saving... - loss_string = "Training [#{0:06d}][{1:.5s}s]".format ( self.epoch, '{:0.4f}'.format(epoch_time) ) + time_str = time.strftime("[%H:%M:%S]") + if iter_time >= 10: + loss_string = "{0}[#{1:06d}][{2:.5s}s]".format ( time_str, self.iter, '{:0.4f}'.format(iter_time) ) else: - loss_string = "Training [#{0:06d}][{1:04d}ms]".format ( self.epoch, int(epoch_time*1000) ) + loss_string = "{0}[#{1:06d}][{2:04d}ms]".format ( time_str, self.iter, int(iter_time*1000) ) for (loss_name, loss_value) in losses: loss_string += " %s:%.3f" % (loss_name, loss_value) return loss_string - def pass_one_epoch(self): + def pass_one_iter(self): self.last_sample = self.generate_next_sample() def finalize(self): nnlib.finalize_all() def is_first_run(self): - return self.epoch == 0 + return self.iter == 0 def is_debug(self): return self.debug @@ -394,8 +398,8 @@ class ModelBase(object): def get_batch_size(self): return self.batch_size - def get_epoch(self): - return self.epoch + def get_iter(self): + return self.iter def get_loss_history(self): return self.loss_history @@ -430,7 +434,7 @@ class ModelBase(object): self.batch_size = d[ keys[-1] ] @staticmethod - def get_loss_history_preview(loss_history, epoch, w, c): + def get_loss_history_preview(loss_history, iter, w, c): loss_history = np.array (loss_history.copy()) lh_height = 100 @@ -483,7 +487,7 @@ class ModelBase(object): last_line_t = int((lh_lines-1)*lh_line_height) last_line_b = int(lh_lines*lh_line_height) - lh_text = 'Epoch: %d' % (epoch) if epoch != 0 else '' + lh_text = 'Iter: %d' % (iter) if iter != 0 else '' lh_img[last_line_t:last_line_b, 0:w] += image_utils.get_text_image ( (w,last_line_b-last_line_t,c), lh_text, color=[0.8]*c ) return lh_img \ No newline at end of file diff --git a/models/Model_DF/Model.py b/models/Model_DF/Model.py index e5f0639..90e91f7 100644 --- a/models/Model_DF/Model.py +++ b/models/Model_DF/Model.py @@ -12,7 +12,7 @@ class Model(ModelBase): def onInitializeOptions(self, is_first_run, ask_override): if is_first_run or ask_override: def_pixel_loss = self.options.get('pixel_loss', False) - self.options['pixel_loss'] = io.input_bool ("Use pixel loss? (y/n, ?:help skip: n/default ) : ", def_pixel_loss, help_message="Default DSSIM loss good for initial understanding structure of faces. Use pixel loss after 20k epochs to enhance fine details and decrease face jitter.") + self.options['pixel_loss'] = io.input_bool ("Use pixel loss? (y/n, ?:help skip: n/default ) : ", def_pixel_loss, help_message="Default DSSIM loss good for initial understanding structure of faces. Use pixel loss after 20k iters to enhance fine details and decrease face jitter.") else: self.options['pixel_loss'] = self.options.get('pixel_loss', False) @@ -62,7 +62,7 @@ class Model(ModelBase): [self.decoder_dst, 'decoder_dst.h5']] ) #override - def onTrainOneEpoch(self, sample, generators_list): + def onTrainOneIter(self, sample, generators_list): warped_src, target_src, target_src_mask = sample[0] warped_dst, target_dst, target_dst_mask = sample[1] diff --git a/models/Model_H128/Model.py b/models/Model_H128/Model.py index 153a808..404bdab 100644 --- a/models/Model_H128/Model.py +++ b/models/Model_H128/Model.py @@ -20,7 +20,7 @@ class Model(ModelBase): if is_first_run or ask_override: def_pixel_loss = self.options.get('pixel_loss', False) - self.options['pixel_loss'] = io.input_bool ("Use pixel loss? (y/n, ?:help skip: n/default ) : ", def_pixel_loss, help_message="Default DSSIM loss good for initial understanding structure of faces. Use pixel loss after 20k epochs to enhance fine details and decrease face jitter.") + self.options['pixel_loss'] = io.input_bool ("Use pixel loss? (y/n, ?:help skip: n/default ) : ", def_pixel_loss, help_message="Default DSSIM loss good for initial understanding structure of faces. Use pixel loss after 20k iters to enhance fine details and decrease face jitter.") else: self.options['pixel_loss'] = self.options.get('pixel_loss', False) @@ -77,7 +77,7 @@ class Model(ModelBase): [self.decoder_dst, 'decoder_dst.h5']] ) #override - def onTrainOneEpoch(self, sample, generators_list): + def onTrainOneIter(self, sample, generators_list): warped_src, target_src, target_src_mask = sample[0] warped_dst, target_dst, target_dst_mask = sample[1] diff --git a/models/Model_H64/Model.py b/models/Model_H64/Model.py index fcd76fe..5bb6075 100644 --- a/models/Model_H64/Model.py +++ b/models/Model_H64/Model.py @@ -20,7 +20,7 @@ class Model(ModelBase): if is_first_run or ask_override: def_pixel_loss = self.options.get('pixel_loss', False) - self.options['pixel_loss'] = io.input_bool ("Use pixel loss? (y/n, ?:help skip: n/default ) : ", def_pixel_loss, help_message="Default DSSIM loss good for initial understanding structure of faces. Use pixel loss after 20k epochs to enhance fine details and decrease face jitter.") + self.options['pixel_loss'] = io.input_bool ("Use pixel loss? (y/n, ?:help skip: n/default ) : ", def_pixel_loss, help_message="Default DSSIM loss good for initial understanding structure of faces. Use pixel loss after 20k iters to enhance fine details and decrease face jitter.") else: self.options['pixel_loss'] = self.options.get('pixel_loss', False) @@ -78,7 +78,7 @@ class Model(ModelBase): [self.decoder_dst, 'decoder_dst.h5']] ) #override - def onTrainOneEpoch(self, sample, generators_list): + def onTrainOneIter(self, sample, generators_list): warped_src, target_src, target_src_full_mask = sample[0] warped_dst, target_dst, target_dst_full_mask = sample[1] diff --git a/models/Model_LIAEF128/Model.py b/models/Model_LIAEF128/Model.py index 1e881bc..91472ef 100644 --- a/models/Model_LIAEF128/Model.py +++ b/models/Model_LIAEF128/Model.py @@ -12,7 +12,7 @@ class Model(ModelBase): def onInitializeOptions(self, is_first_run, ask_override): if is_first_run or ask_override: def_pixel_loss = self.options.get('pixel_loss', False) - self.options['pixel_loss'] = io.input_bool ("Use pixel loss? (y/n, ?:help skip: n/default ) : ", def_pixel_loss, help_message="Default DSSIM loss good for initial understanding structure of faces. Use pixel loss after 20k epochs to enhance fine details and decrease face jitter.") + self.options['pixel_loss'] = io.input_bool ("Use pixel loss? (y/n, ?:help skip: n/default ) : ", def_pixel_loss, help_message="Default DSSIM loss good for initial understanding structure of faces. Use pixel loss after 20k iters to enhance fine details and decrease face jitter.") else: self.options['pixel_loss'] = self.options.get('pixel_loss', False) @@ -70,7 +70,7 @@ class Model(ModelBase): [self.inter_AB, 'inter_AB.h5']] ) #override - def onTrainOneEpoch(self, sample, generators_list): + def onTrainOneIter(self, sample, generators_list): warped_src, target_src, target_src_mask = sample[0] warped_dst, target_dst, target_dst_mask = sample[1] diff --git a/models/Model_SAE/Model.py b/models/Model_SAE/Model.py index 9cc199f..d2151b4 100644 --- a/models/Model_SAE/Model.py +++ b/models/Model_SAE/Model.py @@ -77,11 +77,11 @@ class SAEModel(ModelBase): default_bg_style_power = 0.0 if is_first_run or ask_override: def_pixel_loss = self.options.get('pixel_loss', False) - self.options['pixel_loss'] = io.input_bool ("Use pixel loss? (y/n, ?:help skip: %s ) : " % (yn_str[def_pixel_loss]), def_pixel_loss, help_message="Default DSSIM loss good for initial understanding structure of faces. Use pixel loss after 15-25k epochs to enhance fine details and decrease face jitter.") + self.options['pixel_loss'] = io.input_bool ("Use pixel loss? (y/n, ?:help skip: %s ) : " % (yn_str[def_pixel_loss]), def_pixel_loss, help_message="Default DSSIM loss good for initial understanding structure of faces. Use pixel loss after 15-25k iters to enhance fine details and decrease face jitter.") default_face_style_power = default_face_style_power if is_first_run else self.options.get('face_style_power', default_face_style_power) self.options['face_style_power'] = np.clip ( io.input_number("Face style power ( 0.0 .. 100.0 ?:help skip:%.2f) : " % (default_face_style_power), default_face_style_power, - help_message="Learn to transfer face style details such as light and color conditions. Warning: Enable it only after 10k epochs, when predicted face is clear enough to start learn style. Start from 0.1 value and check history changes."), 0.0, 100.0 ) + help_message="Learn to transfer face style details such as light and color conditions. Warning: Enable it only after 10k iters, when predicted face is clear enough to start learn style. Start from 0.1 value and check history changes."), 0.0, 100.0 ) default_bg_style_power = default_bg_style_power if is_first_run else self.options.get('bg_style_power', default_bg_style_power) self.options['bg_style_power'] = np.clip ( io.input_number("Background style power ( 0.0 .. 100.0 ?:help skip:%.2f) : " % (default_bg_style_power), default_bg_style_power, @@ -107,7 +107,6 @@ class SAEModel(ModelBase): masked_training = True - epoch_alpha = Input( (1,) ) warped_src = Input(bgr_shape) target_src = Input(bgr_shape) target_srcm = Input(mask_shape) @@ -395,7 +394,7 @@ class SAEModel(ModelBase): #override - def onTrainOneEpoch(self, generators_samples, generators_list): + def onTrainOneIter(self, generators_samples, generators_list): src_samples = generators_samples[0] dst_samples = generators_samples[1]