mirror of
https://github.com/deepfakes/faceswap
synced 2025-06-07 10:43:27 -04:00
* model_refactor (#571) * original model to new structure * IAE model to new structure * OriginalHiRes to new structure * Fix trainer for different resolutions * Initial config implementation * Configparse library added * improved training data loader * dfaker model working * Add logging to training functions * Non blocking input for cli training * Add error handling to threads. Add non-mp queues to queue_handler * Improved Model Building and NNMeta * refactor lib/models * training refactor. DFL H128 model Implementation * Dfaker - use hashes * Move timelapse. Remove perceptual loss arg * Update INSTALL.md. Add logger formatting. Update Dfaker training * DFL h128 partially ported * Add mask to dfaker (#573) * Remove old models. Add mask to dfaker * dfl mask. Make masks selectable in config (#575) * DFL H128 Mask. Mask type selectable in config. * remove gan_v2_2 * Creating Input Size config for models Creating Input Size config for models Will be used downstream in converters. Also name change of image_shape to input_shape to clarify ( for future models with potentially different output_shapes) * Add mask loss options to config * MTCNN options to config.ini. Remove GAN config. Update USAGE.md * Add sliders for numerical values in GUI * Add config plugins menu to gui. Validate config * Only backup model if loss has dropped. Get training working again * bugfixes * Standardise loss printing * GUI idle cpu fixes. Graph loss fix. * mutli-gpu logging bugfix * Merge branch 'staging' into train_refactor * backup state file * Crash protection: Only backup if both total losses have dropped * Port OriginalHiRes_RC4 to train_refactor (OriginalHiRes) * Load and save model structure with weights * Slight code update * Improve config loader. Add subpixel opt to all models. Config to state * Show samples... wrong input * Remove AE topology. Add input/output shapes to State * Port original_villain (birb/VillainGuy) model to faceswap * Add plugin info to GUI config pages * Load input shape from state. IAE Config options. * Fix transform_kwargs. Coverage to ratio. Bugfix mask detection * Suppress keras userwarnings. Automate zoom. Coverage_ratio to model def. * Consolidation of converters & refactor (#574) * Consolidation of converters & refactor Initial Upload of alpha Items - consolidate convert_mased & convert_adjust into one converter -add average color adjust to convert_masked -allow mask transition blur size to be a fixed integer of pixels and a fraction of the facial mask size -allow erosion/dilation size to be a fixed integer of pixels and a fraction of the facial mask size -eliminate redundant type conversions to avoid multiple round-off errors -refactor loops for vectorization/speed -reorganize for clarity & style changes TODO - bug/issues with warping the new face onto a transparent old image...use a cleanup mask for now - issues with mask border giving black ring at zero erosion .. investigate - remove GAN ?? - test enlargment factors of umeyama standard face .. match to coverage factor - make enlargment factor a model parameter - remove convert_adjusted and referencing code when finished * Update Convert_Masked.py default blur size of 2 to match original... description of enlargement tests breakout matrxi scaling into def * Enlargment scale as a cli parameter * Update cli.py * dynamic interpolation algorithm Compute x & y scale factors from the affine matrix on the fly by QR decomp. Choose interpolation alogrithm for the affine warp based on an upsample or downsample for each image * input size input size from config * fix issues with <1.0 erosion * Update convert.py * Update Convert_Adjust.py more work on the way to merginf * Clean up help note on sharpen * cleanup seamless * Delete Convert_Adjust.py * Update umeyama.py * Update training_data.py * swapping * segmentation stub * changes to convert.str * Update masked.py * Backwards compatibility fix for models Get converter running * Convert: Move masks to class. bugfix blur_size some linting * mask fix * convert fixes - missing facehull_rect re-added - coverage to % - corrected coverage logic - cleanup of gui option ordering * Update cli.py * default for blur * Update masked.py * added preliminary low_mem version of OriginalHighRes model plugin * Code cleanup, minor fixes * Update masked.py * Update masked.py * Add dfl mask to convert * histogram fix & seamless location * update * revert * bugfix: Load actual configuration in gui * Standardize nn_blocks * Update cli.py * Minor code amends * Fix Original HiRes model * Add masks to preview output for mask trainers refactor trainer.__base.py * Masked trainers converter support * convert bugfix * Bugfix: Converter for masked (dfl/dfaker) trainers * Additional Losses (#592) * initial upload * Delete blur.py * default initializer = He instead of Glorot (#588) * Allow kernel_initializer to be overridable * Add ICNR Initializer option for upscale on all models. * Hopefully fixes RSoDs with original-highres model plugin * remove debug line * Original-HighRes model plugin Red Screen of Death fix, take #2 * Move global options to _base. Rename Villain model * clipnorm and res block biases * scale the end of res block * res block * dfaker pre-activation res * OHRES pre-activation * villain pre-activation * tabs/space in nn_blocks * fix for histogram with mask all set to zero * fix to prevent two networks with same name * GUI: Wider tooltips. Improve TQDM capture * Fix regex bug * Convert padding=48 to ratio of image size * Add size option to alignments tool extract * Pass through training image size to convert from model * Convert: Pull training coverage from model * convert: coverage, blur and erode to percent * simplify matrix scaling * ordering of sliders in train * Add matrix scaling to utils. Use interpolation in lib.aligner transform * masked.py Import get_matrix_scaling from utils * fix circular import * Update masked.py * quick fix for matrix scaling * testing thus for now * tqdm regex capture bugfix * Minor ammends * blur size cleanup * Remove coverage option from convert (Now cascades from model) * Implement convert for all model types * Add mask option and coverage option to all existing models * bugfix for model loading on convert * debug print removal * Bugfix for masks in dfl_h128 and iae * Update preview display. Add preview scaling to cli * mask notes * Delete training_data_v2.py errant file * training data variables * Fix timelapse function * Add new config items to state file for legacy purposes * Slight GUI tweak * Raise exception if problem with loaded model * Add Tensorboard support (Logs stored in model directory) * ICNR fix * loss bugfix * convert bugfix * Move ini files to config folder. Make TensorBoard optional * Fix training data for unbalanced inputs/outputs * Fix config "none" test * Keep helptext in .ini files when saving config from GUI * Remove frame_dims from alignments * Add no-flip and warp-to-landmarks cli options * Revert OHR to RC4_fix version * Fix lowmem mode on OHR model * padding to variable * Save models in parallel threads * Speed-up of res_block stability * Automated Reflection Padding * Reflect Padding as a training option Includes auto-calculation of proper padding shapes, input_shapes, output_shapes Flag included in config now * rest of reflect padding * Move TB logging to cli. Session info to state file * Add session iterations to state file * Add recent files to menu. GUI code tidy up * [GUI] Fix recent file list update issue * Add correct loss names to TensorBoard logs * Update live graph to use TensorBoard and remove animation * Fix analysis tab. GUI optimizations * Analysis Graph popup to Tensorboard Logs * [GUI] Bug fix for graphing for models with hypens in name * [GUI] Correctly split loss to tabs during training * [GUI] Add loss type selection to analysis graph * Fix store command name in recent files. Switch to correct tab on open * [GUI] Disable training graph when 'no-logs' is selected * Fix graphing race condition * rename original_hires model to unbalanced
323 lines
12 KiB
Python
Executable file
323 lines
12 KiB
Python
Executable file
#!/usr/bin python3
|
|
""" Graph functions for Display Frame of the Faceswap GUI """
|
|
import datetime
|
|
import logging
|
|
import os
|
|
import tkinter as tk
|
|
|
|
from tkinter import ttk
|
|
from math import ceil, floor
|
|
|
|
import matplotlib
|
|
# pylint: disable=wrong-import-position
|
|
matplotlib.use("TkAgg")
|
|
|
|
from matplotlib import pyplot as plt, style # noqa
|
|
from matplotlib.backends.backend_tkagg import (FigureCanvasTkAgg,
|
|
NavigationToolbar2Tk) # noqa
|
|
|
|
from .tooltip import Tooltip # noqa
|
|
from .utils import get_config, get_images # noqa
|
|
|
|
logger = logging.getLogger(__name__) # pylint: disable=invalid-name
|
|
|
|
|
|
class NavigationToolbar(NavigationToolbar2Tk): # pylint: disable=too-many-ancestors
|
|
""" Same as default, but only including buttons we need
|
|
with custom icons and layout
|
|
From: https://stackoverflow.com/questions/12695678 """
|
|
toolitems = [t for t in NavigationToolbar2Tk.toolitems if
|
|
t[0] in ("Home", "Pan", "Zoom", "Save")]
|
|
|
|
@staticmethod
|
|
def _Button(frame, text, file, command, extension=".gif"): # pylint: disable=arguments-differ
|
|
""" Map Buttons to their own frame.
|
|
Use custom button icons, Use ttk buttons pack to the right """
|
|
iconmapping = {"home": "reset",
|
|
"filesave": "save",
|
|
"zoom_to_rect": "zoom"}
|
|
icon = iconmapping[file] if iconmapping.get(file, None) else file
|
|
img = get_images().icons[icon]
|
|
btn = ttk.Button(frame, text=text, image=img, command=command)
|
|
btn.pack(side=tk.RIGHT, padx=2)
|
|
return btn
|
|
|
|
def _init_toolbar(self):
|
|
""" Same as original but ttk widgets and standard tooltips used. Separator added and
|
|
message label packed to the left """
|
|
xmin, xmax = self.canvas.figure.bbox.intervalx
|
|
height, width = 50, xmax-xmin
|
|
ttk.Frame.__init__(self, master=self.window, width=int(width), height=int(height))
|
|
|
|
sep = ttk.Frame(self, height=2, relief=tk.RIDGE)
|
|
sep.pack(fill=tk.X, pady=(5, 0), side=tk.TOP)
|
|
|
|
self.update() # Make axes menu
|
|
|
|
btnframe = ttk.Frame(self)
|
|
btnframe.pack(fill=tk.X, padx=5, pady=5, side=tk.RIGHT)
|
|
|
|
for text, tooltip_text, image_file, callback in self.toolitems:
|
|
if text is None:
|
|
# Add a spacer; return value is unused.
|
|
self._Spacer()
|
|
else:
|
|
button = self._Button(btnframe, text=text, file=image_file,
|
|
command=getattr(self, callback))
|
|
if tooltip_text is not None:
|
|
Tooltip(button, text=tooltip_text, wraplength=200)
|
|
|
|
self.message = tk.StringVar(master=self)
|
|
self._message_label = ttk.Label(master=self, textvariable=self.message)
|
|
self._message_label.pack(side=tk.LEFT, padx=5)
|
|
self.pack(side=tk.BOTTOM, fill=tk.X)
|
|
|
|
|
|
class GraphBase(ttk.Frame): # pylint: disable=too-many-ancestors
|
|
""" Base class for matplotlib line graphs """
|
|
def __init__(self, parent, data, ylabel):
|
|
logger.debug("Initializing %s", self.__class__.__name__)
|
|
super().__init__(parent)
|
|
style.use("ggplot")
|
|
|
|
self.calcs = data
|
|
self.ylabel = ylabel
|
|
self.colourmaps = ["Reds", "Blues", "Greens", "Purples", "Oranges",
|
|
"Greys", "copper", "summer", "bone"]
|
|
self.lines = list()
|
|
self.toolbar = None
|
|
self.fig = plt.figure(figsize=(4, 4), dpi=75)
|
|
self.ax1 = self.fig.add_subplot(1, 1, 1)
|
|
self.plotcanvas = FigureCanvasTkAgg(self.fig, self)
|
|
|
|
self.initiate_graph()
|
|
self.update_plot(initiate=True)
|
|
logger.debug("Initialized %s", self.__class__.__name__)
|
|
|
|
def initiate_graph(self):
|
|
""" Place the graph canvas """
|
|
logger.debug("Setting plotcanvas")
|
|
self.plotcanvas.get_tk_widget().pack(side=tk.TOP, padx=5, fill=tk.BOTH, expand=True)
|
|
plt.subplots_adjust(left=0.100,
|
|
bottom=0.100,
|
|
right=0.95,
|
|
top=0.95,
|
|
wspace=0.2,
|
|
hspace=0.2)
|
|
logger.debug("Set plotcanvas")
|
|
|
|
def update_plot(self, initiate=True):
|
|
""" Update the plot with incoming data """
|
|
logger.trace("Updating plot")
|
|
if initiate:
|
|
logger.debug("Initializing plot")
|
|
self.lines = list()
|
|
self.ax1.clear()
|
|
self.axes_labels_set()
|
|
logger.debug("Initialized plot")
|
|
|
|
fulldata = [item for item in self.calcs.stats.values()]
|
|
self.axes_limits_set(fulldata)
|
|
|
|
xrng = [x for x in range(self.calcs.iterations)]
|
|
keys = list(self.calcs.stats.keys())
|
|
for idx, item in enumerate(self.lines_sort(keys)):
|
|
if initiate:
|
|
self.lines.extend(self.ax1.plot(xrng, self.calcs.stats[item[0]],
|
|
label=item[1], linewidth=item[2], color=item[3]))
|
|
else:
|
|
self.lines[idx].set_data(xrng, self.calcs.stats[item[0]])
|
|
|
|
if initiate:
|
|
self.legend_place()
|
|
logger.trace("Updated plot")
|
|
|
|
def axes_labels_set(self):
|
|
""" Set the axes label and range """
|
|
logger.debug("Setting axes labels. y-label: '%s'", self.ylabel)
|
|
self.ax1.set_xlabel("Iterations")
|
|
self.ax1.set_ylabel(self.ylabel)
|
|
|
|
def axes_limits_set_default(self):
|
|
""" Set default axes limits """
|
|
logger.debug("Setting default axes ranges")
|
|
self.ax1.set_ylim(0.00, 100.0)
|
|
self.ax1.set_xlim(0, 1)
|
|
|
|
def axes_limits_set(self, data):
|
|
""" Set the axes limits """
|
|
xmax = self.calcs.iterations - 1 if self.calcs.iterations > 1 else 1
|
|
if data:
|
|
ymin, ymax = self.axes_data_get_min_max(data)
|
|
self.ax1.set_ylim(ymin, ymax)
|
|
self.ax1.set_xlim(0, xmax)
|
|
else:
|
|
self.axes_limits_set_default()
|
|
logger.trace("axes ranges: (y: (%s, %s), x:(0, %s)", ymin, ymax, xmax)
|
|
|
|
@staticmethod
|
|
def axes_data_get_min_max(data):
|
|
""" Return the minimum and maximum values from list of lists """
|
|
ymin, ymax = list(), list()
|
|
for item in data:
|
|
dataset = list(filter(lambda x: x is not None, item))
|
|
if not dataset:
|
|
continue
|
|
ymin.append(min(dataset) * 1000)
|
|
ymax.append(max(dataset) * 1000)
|
|
ymin = floor(min(ymin)) / 1000
|
|
ymax = ceil(max(ymax)) / 1000
|
|
logger.trace("ymin: %s, ymax: %s", ymin, ymax)
|
|
return ymin, ymax
|
|
|
|
def axes_set_yscale(self, scale):
|
|
""" Set the Y-Scale to log or linear """
|
|
logger.debug("yscale: '%s'", scale)
|
|
self.ax1.set_yscale(scale)
|
|
|
|
def lines_sort(self, keys):
|
|
""" Sort the data keys into consistent order
|
|
and set line color map and line width """
|
|
logger.trace("Sorting lines")
|
|
raw_lines = list()
|
|
sorted_lines = list()
|
|
for key in sorted(keys):
|
|
title = key.replace("_", " ").title()
|
|
if key.startswith(("avg", "trend")):
|
|
sorted_lines.append([key, title])
|
|
else:
|
|
raw_lines.append([key, title])
|
|
|
|
groupsize = self.lines_groupsize(raw_lines, sorted_lines)
|
|
sorted_lines = raw_lines + sorted_lines
|
|
lines = self.lines_style(sorted_lines, groupsize)
|
|
return lines
|
|
|
|
@staticmethod
|
|
def lines_groupsize(raw_lines, sorted_lines):
|
|
""" Get the number of items in each group.
|
|
If raw data isn't selected, then check the length of
|
|
remaining groups until something is found """
|
|
groupsize = 1
|
|
if raw_lines:
|
|
groupsize = len(raw_lines)
|
|
else:
|
|
for check in ("avg", "trend"):
|
|
if any(item[0].startswith(check) for item in sorted_lines):
|
|
groupsize = len([item for item in sorted_lines if item[0].startswith(check)])
|
|
break
|
|
logger.trace(groupsize)
|
|
return groupsize
|
|
|
|
def lines_style(self, lines, groupsize):
|
|
""" Set the color map and line width for each group """
|
|
logger.trace("Setting lines style")
|
|
groups = int(len(lines) / groupsize)
|
|
colours = self.lines_create_colors(groupsize, groups)
|
|
for idx, item in enumerate(lines):
|
|
linewidth = ceil((idx + 1) / groupsize)
|
|
item.extend((linewidth, colours[idx]))
|
|
return lines
|
|
|
|
def lines_create_colors(self, groupsize, groups):
|
|
""" Create the colors """
|
|
colours = list()
|
|
for i in range(1, groups + 1):
|
|
for colour in self.colourmaps[0:groupsize]:
|
|
cmap = matplotlib.cm.get_cmap(colour)
|
|
cpoint = 1 - (i / 5)
|
|
colours.append(cmap(cpoint))
|
|
logger.trace(colours)
|
|
return colours
|
|
|
|
def legend_place(self):
|
|
""" Place and format legend """
|
|
logger.debug("Placing legend")
|
|
self.ax1.legend(loc="upper right", ncol=2)
|
|
|
|
def toolbar_place(self, parent):
|
|
""" Add Graph Navigation toolbar """
|
|
logger.debug("Placing toolbar")
|
|
self.toolbar = NavigationToolbar(self.plotcanvas, parent)
|
|
self.toolbar.pack(side=tk.BOTTOM)
|
|
self.toolbar.update()
|
|
|
|
|
|
class TrainingGraph(GraphBase): # pylint: disable=too-many-ancestors
|
|
""" Live graph to be displayed during training. """
|
|
|
|
def __init__(self, parent, data, ylabel):
|
|
GraphBase.__init__(self, parent, data, ylabel)
|
|
self.add_callback()
|
|
|
|
def add_callback(self):
|
|
""" Add the variable trace to update graph on recent button or save iteration """
|
|
get_config().tk_vars["refreshgraph"].trace("w", self.refresh)
|
|
|
|
def build(self):
|
|
""" Update the plot area with loss values """
|
|
logger.debug("Building training graph")
|
|
self.plotcanvas.draw()
|
|
logger.debug("Built training graph")
|
|
|
|
def refresh(self, *args): # pylint: disable=unused-argument
|
|
""" Read loss data and apply to graph """
|
|
logger.debug("Updating plot")
|
|
self.calcs.refresh()
|
|
self.update_plot(initiate=False)
|
|
self.plotcanvas.draw()
|
|
get_config().tk_vars["refreshgraph"].set(False)
|
|
|
|
def save_fig(self, location):
|
|
""" Save the figure to file """
|
|
logger.debug("Saving graph: '%s'", location)
|
|
keys = sorted([key.replace("raw_", "") for key in self.calcs.stats.keys()
|
|
if key.startswith("raw_")])
|
|
filename = " - ".join(keys)
|
|
now = datetime.datetime.now().strftime("%Y%m%d_%H%M%S")
|
|
filename = os.path.join(location, "{}_{}.{}".format(filename, now, "png"))
|
|
self.fig.set_size_inches(16, 9)
|
|
self.fig.savefig(filename, bbox_inches="tight", dpi=120)
|
|
print("Saved graph to {}".format(filename))
|
|
logger.debug("Saved graph: '%s'", filename)
|
|
self.resize_fig()
|
|
|
|
def resize_fig(self):
|
|
""" Resize the figure back to the canvas """
|
|
class Event(): # pylint: disable=too-few-public-methods
|
|
""" Event class that needs to be passed to plotcanvas.resize """
|
|
pass
|
|
Event.width = self.winfo_width()
|
|
Event.height = self.winfo_height()
|
|
self.plotcanvas.resize(Event) # pylint: disable=no-value-for-parameter
|
|
|
|
|
|
class SessionGraph(GraphBase): # pylint: disable=too-many-ancestors
|
|
""" Session Graph for session pop-up """
|
|
def __init__(self, parent, data, ylabel, scale):
|
|
GraphBase.__init__(self, parent, data, ylabel)
|
|
self.scale = scale
|
|
|
|
def build(self):
|
|
""" Build the session graph """
|
|
logger.debug("Building session graph")
|
|
self.toolbar_place(self)
|
|
self.plotcanvas.draw()
|
|
logger.debug("Built session graph")
|
|
|
|
def refresh(self, data, ylabel, scale):
|
|
""" Refresh graph data """
|
|
logger.debug("Refreshing session graph: (ylabel: '%s', scale: '%s')", ylabel, scale)
|
|
self.calcs = data
|
|
self.ylabel = ylabel
|
|
self.set_yscale_type(scale)
|
|
logger.debug("Refreshed session graph")
|
|
|
|
def set_yscale_type(self, scale):
|
|
""" switch the y-scale and redraw """
|
|
logger.debug("Updating scale type: '%s'", scale)
|
|
self.scale = scale
|
|
self.update_plot(initiate=True)
|
|
self.axes_set_yscale(self.scale)
|
|
self.plotcanvas.draw()
|
|
logger.debug("Updated scale type")
|