train_adversarial.py

# YOLOv5 🚀 by Ultralytics, GPL-3.0 license
"""
Train a YOLOv5 model on a custom dataset

Usage:
    $ python path/to/train.py --data coco128.yaml --weights yolov5s.pt --img 640
"""

import argparse
import logging
import math
import os
import random
import sys
import time
from copy import deepcopy
from datetime import datetime
from pathlib import Path

import numpy as np
import torch
import torch.distributed as dist
import torch.nn as nn
import yaml
from torch.cuda import amp
from torch.nn.parallel import DistributedDataParallel as DDP
from torch.optim import SGD, Adam, lr_scheduler
from tqdm import tqdm

FILE = Path(__file__).resolve()
ROOT = FILE.parents[0]  # YOLOv5 root directory
if str(ROOT) not in sys.path:
    sys.path.append(str(ROOT))  # add ROOT to PATH
ROOT = Path(os.path.relpath(ROOT, Path.cwd()))  # relative

import val, reconstruct  # for end-of-epoch mAP
from models.experimental import attempt_load
from models.supplemental import AutoEncoder, Decoder_Rec
from models.yolo import Model
from utils.autoanchor import check_anchors
from utils.autobatch import check_train_batch_size
from utils.datasets import create_dataloader
from utils.general import (LOGGER, check_dataset, check_file, check_git_status, check_img_size, check_requirements,
                           check_suffix, check_yaml, colorstr, get_latest_run, increment_path, init_seeds,
                           intersect_dicts, labels_to_class_weights, labels_to_image_weights, methods, one_cycle,
                           print_args, print_mutation, strip_optimizer, print_args_to_file, make_divisible)
from utils.downloads import attempt_download
from utils.loss import ComputeLoss, CompressibilityLoss, ComputeRecLoss
from utils.plots import plot_labels, plot_evolve
from utils.torch_utils import EarlyStopping, ModelEMA, de_parallel, select_device, torch_distributed_zero_first
from utils.loggers.wandb.wandb_utils import check_wandb_resume
from utils.metrics import fitness_adv, fitness
from utils.loggers import Loggers
from utils.callbacks import Callbacks

LOCAL_RANK = int(os.getenv('LOCAL_RANK', -1))  # https://pytorch.org/docs/stable/elastic/run.html
RANK = int(os.getenv('RANK', -1))
WORLD_SIZE = int(os.getenv('WORLD_SIZE', 1))

print(LOCAL_RANK, RANK, WORLD_SIZE)

def train(hyp,  # path/to/hyp.yaml or hyp dictionary
          opt,
          device,
          callbacks
          ):
    save_dir, epochs, batch_size, weights, single_cls, evolve, data, cfg, resume, noval, nosave, workers, freeze, train_yolo, freeze_autoenc, cutting_layer, autoenc_chs= \
        Path(opt.save_dir), opt.epochs, opt.batch_size, opt.weights, opt.single_cls, opt.evolve, opt.data, opt.cfg, \
        opt.resume, opt.noval, opt.nosave, opt.workers, opt.freeze, opt.train_yolo, opt.freeze_autoenc, opt.cut_layer, opt.autoenc_chs

    # Directories
    w = save_dir / 'weights'  # weights dir
    (w.parent if evolve else w).mkdir(parents=True, exist_ok=True)  # make dir
    last, best = w / 'last.pt', w / 'best.pt'
    rec_dir = (save_dir / 'reconstructions')
    rec_dir.mkdir(parents=True, exist_ok=True)

    # Hyperparameters
    if isinstance(hyp, str):
        with open(hyp, errors='ignore') as f:
            hyp = yaml.safe_load(f)  # load hyps dict
    hyp['lr0'] = opt.lr0
    hyp['lrf'] = opt.lrf
    hyp['warmup_bias_lr'] = opt.warmup_bias_lr
    hyp['weight_decay'] = opt.weight_decay
    LOGGER.info(colorstr('hyperparameters: ') + ', '.join(f'{k}={v}' for k, v in hyp.items()))

    data_dict = None
    # Loggers
    if RANK in [-1, 0]:
        loggers = Loggers(save_dir, weights, opt, hyp, LOGGER)  # loggers instance
        if loggers.wandb:
            data_dict = loggers.wandb.data_dict
            if resume:
                weights, epochs, hyp = opt.weights, opt.epochs, opt.hyp

        # Register actions
        for k in methods(loggers):
            callbacks.register_action(k, callback=getattr(loggers, k))

    # Config
    plots = not evolve  # create plots
    cuda = device.type != 'cpu'
    init_seeds(1 + RANK)
    with torch_distributed_zero_first(LOCAL_RANK):
        data_dict = data_dict or check_dataset(data)  # check if None
    train_path, val_path = data_dict['train'], data_dict['val']
    nc = 1 if single_cls else int(data_dict['nc'])  # number of classes
    names = ['item'] if single_cls and len(data_dict['names']) != 1 else data_dict['names']  # class names
    assert len(names) == nc, f'{len(names)} names found for nc={nc} dataset in {data}'  # check
    is_coco = isinstance(val_path, str) and val_path.endswith('coco/val2017.txt')  # COCO dataset

    # Models 
    check_suffix(weights, '.pt')  # check weights
    pretrained = weights.endswith('.pt')
    if pretrained:
        # ---- Object Detection Model -----#
        with torch_distributed_zero_first(LOCAL_RANK):
            weights = attempt_download(weights)  # download if not found locally
        ckpt = torch.load(weights, map_location=device)  # load checkpoint
        cutting_layer = getattr(ckpt['model'], 'cutting_layer', opt.cut_layer)
        model = Model(cfg or ckpt['model'].yaml, ch=3, nc=nc, anchors=hyp.get('anchors'), cutting_layer=cutting_layer).to(device)  # create
        exclude = ['anchor'] if (cfg or hyp.get('anchors')) and not resume else []  # exclude keys
        csd = ckpt['model'].float().state_dict()  # checkpoint state_dict as FP32
        csd = intersect_dicts(csd, model.state_dict(), exclude=exclude)  # intersect
        model.load_state_dict(csd, strict=False)  # load
        LOGGER.info(f'Transferred {len(csd)}/{len(model.state_dict())} items from {weights}')  # report
        # ---- Autoencoder -----#
        if 'autoencoder' in ckpt:
            autoenc_chs = ckpt['autoencoder'].chs
            autoencoder = AutoEncoder(autoenc_chs).to(device)
            autoencoder.load_state_dict(ckpt['autoencoder'].float().state_dict())
            print('pretrained autoencoder')
        else:
            autoencoder = AutoEncoder(autoenc_chs).to(device)
        # ---- Reconstruction Model -----#
        if 'rec_model' in ckpt and not opt.rec_only:
            rec_model = Decoder_Rec(cin=ckpt['rec_model'].cin, cout=ckpt['rec_model'].cout, first_chs=autoenc_chs).to(device)
            rec_model.load_state_dict(ckpt['rec_model'].float().state_dict())
            print('pretrained reconstruction model')
        else:
            rec_model = Decoder_Rec(cin=make_divisible((model.yaml['backbone'][cutting_layer][3][0]) * model.yaml['width_multiple'], 8), cout=3, first_chs=autoenc_chs).to(device)
    else:
        model = Model(cfg, ch=3, nc=nc, anchors=hyp.get('anchors'), cutting_layer=cutting_layer).to(device)  # create
        autoencoder = AutoEncoder(autoenc_chs).to(device)
        rec_model = Decoder_Rec(cin=make_divisible((model.yaml['backbone'][cutting_layer][3][0]) * model.yaml['width_multiple'], 8), cout=3, first_chs=autoenc_chs).to(device)
    LOGGER.info(f'autoencoder channels: {autoenc_chs}')

    # Save run settings
    opt.autoenc_chs = autoenc_chs
    opt.cut_layer = cutting_layer
    with open(save_dir / 'hyp.yaml', 'w') as f:
        yaml.safe_dump(hyp, f, sort_keys=False)
    with open(save_dir / 'opt.yaml', 'w') as f:
        yaml.safe_dump(vars(opt), f, sort_keys=False)
    print_args_to_file(opt, save_dir / 'result.txt')

    if freeze_autoenc:
        for param in autoencoder.parameters():
            param.requires_grad = False
    
    if freeze:
        freeze = freeze
    elif(train_yolo=='all'):
        freeze = 0
    elif(train_yolo=='backend'):
        freeze = 5
    elif(train_yolo=='nothing'):
        freeze = 34 if (pretrained and opt.weights[-4]=='6') else 25
    else:
        raise Exception(f'train-yolo={train_yolo} is not supported')
    # Freeze
    LOGGER.info('\n')
    freeze = [f'model.{x}' for x in range(freeze)]  # layers to freeze
    for k, v in model.named_parameters():
        v.requires_grad = True  # train all layers
        if any(f'{x}.' in k for x in freeze):
            LOGGER.info(f'freezing {k}')
            v.requires_grad = False
    if(train_yolo=='nothing'):  # the last layer(detection) should not be in the eval mode 
        freeze = freeze[:-1]

    # Image size
    gs = max(int(model.stride.max()), 32)  # grid size (max stride)
    imgsz = check_img_size(opt.imgsz, gs, floor=gs * 2)  # verify imgsz is gs-multiple

    # Batch size
    if RANK == -1 and batch_size == -1:  # single-GPU only, estimate best batch size
        batch_size = check_train_batch_size(model, imgsz)

    # Optimizer
    nbs = 64  # nominal batch size
    accumulate = max(round(nbs / batch_size), 1)  # accumulate loss before optimizing
    hyp['weight_decay'] *= batch_size * accumulate / nbs  # scale weight_decay
    LOGGER.info(f"\nScaled weight_decay = {hyp['weight_decay']}")

    g0, g1, g2 = [], [], []  # optimizer parameter groups
    for v in model.modules():
        # print(v)
        if hasattr(v, 'bias') and isinstance(v.bias, nn.Parameter):  # bias
            g2.append(v.bias)
        if isinstance(v, nn.BatchNorm2d):  # weight (no decay)
            g0.append(v.weight)
        elif hasattr(v, 'weight') and isinstance(v.weight, nn.Parameter):  # weight (with decay)
            g1.append(v.weight)
    for v in autoencoder.modules():
        # print(v)
        if hasattr(v, 'bias') and isinstance(v.bias, nn.Parameter):  # bias
            g2.append(v.bias)
        if isinstance(v, nn.BatchNorm2d):  # weight (no decay)
            g0.append(v.weight)
        elif hasattr(v, 'weight') and isinstance(v.weight, nn.Parameter):  # weight (with decay)
            g1.append(v.weight)

    if opt.adam:
        optimizer = Adam(g0, lr=hyp['lr0'], betas=(hyp['momentum'], 0.999))  # adjust beta1 to momentum
    else:
        optimizer = SGD(g0, lr=hyp['lr0'], momentum=hyp['momentum'], nesterov=True)

    optimizer.add_param_group({'params': g1, 'weight_decay': hyp['weight_decay']})  # add g1 with weight_decay
    optimizer.add_param_group({'params': g2})  # add g2 (biases)
    LOGGER.info(f"{colorstr('optimizer:')} {type(optimizer).__name__} with parameter groups "
                f"{len(g0)} weight, {len(g1)} weight (no decay), {len(g2)} bias")
    del g0, g1, g2

    if opt.adam:
        rec_optimizer = Adam(rec_model.parameters(), lr=hyp['lr0'], betas=(hyp['momentum'], 0.999), weight_decay=hyp['weight_decay'])
    else:
        rec_optimizer = SGD(rec_model.parameters(), lr=hyp['lr0'], momentum=hyp['momentum'], nesterov=True, weight_decay=hyp['weight_decay'])

    # Scheduler
    if opt.linear_lr:
        lf = lambda x: (1 - x / (epochs - 1)) * (1.0 - hyp['lrf']) + hyp['lrf']  # linear
    else:
        lf = one_cycle(1, hyp['lrf'], epochs)  # cosine 1->hyp['lrf']
    scheduler = lr_scheduler.LambdaLR(optimizer, lr_lambda=lf)  # plot_lr_scheduler(optimizer, scheduler, epochs)
    rec_scheduler = lr_scheduler.LambdaLR(rec_optimizer, lr_lambda=lf)  # plot_lr_scheduler(optimizer, scheduler, epochs)

    # EMA
    ema = ModelEMA(model) if RANK in [-1, 0] else None

    # Resume
    start_epoch, best_fitness, best_epoch = 0, [float('-inf')], -1
    if pretrained:
        # Optimizer
        if ckpt['optimizer'] is not None:
            optimizer.load_state_dict(ckpt['optimizer'])
        if resume:
            best_fitness = ckpt['best_fitness']

        # EMA
        if ema and ckpt.get('ema'):
            ema.ema.load_state_dict(ckpt['ema'].float().state_dict())
            ema.updates = ckpt['updates']

        # Epochs
        if resume:
            start_epoch = ckpt['epoch'] + 1
            assert start_epoch > 0, f'{weights} training to {epochs} epochs is finished, nothing to resume.'
        if epochs < start_epoch:
            LOGGER.info(f"{weights} has been trained for {ckpt['epoch']} epochs. Fine-tuning for {epochs} more epochs.")
            epochs += ckpt['epoch']  # finetune additional epochs

        del ckpt, csd

    # DP mode
    if cuda and RANK == -1 and torch.cuda.device_count() > 1:
        logging.warning('DP not recommended, instead use torch.distributed.run for best DDP Multi-GPU results.\n'
                        'See Multi-GPU Tutorial at https://github.com/ultralytics/yolov5/issues/475 to get started.')
        model = torch.nn.DataParallel(model)

    # SyncBatchNorm
    if opt.sync_bn and cuda and RANK != -1:
        model = torch.nn.SyncBatchNorm.convert_sync_batchnorm(model).to(device)
        LOGGER.info('Using SyncBatchNorm()')

    # Trainloader
    train_loader, dataset = create_dataloader(train_path, imgsz, batch_size // WORLD_SIZE, gs, single_cls,
                                              hyp=hyp, augment=True, cache=opt.cache, rect=opt.rect, rank=LOCAL_RANK,
                                              workers=workers, image_weights=opt.image_weights, quad=opt.quad,
                                              prefix=colorstr('train: '))
    mlc = int(np.concatenate(dataset.labels, 0)[:, 0].max())  # max label class
    nb = len(train_loader)  # number of batches
    assert mlc < nc, f'Label class {mlc} exceeds nc={nc} in {data}. Possible class labels are 0-{nc - 1}'

    # Process 0
    if RANK in [-1, 0]:
        val_loader = create_dataloader(val_path, imgsz, batch_size // WORLD_SIZE * 2, gs, single_cls,
                                       hyp=hyp, cache=None if noval else opt.cache, rect=True, rank=-1,
                                       workers=workers, pad=0.0,
                                       prefix=colorstr('val: '))[0]

        if not resume:
            labels = np.concatenate(dataset.labels, 0)
            if plots:
                plot_labels(labels, names, save_dir)

            # Anchors
            if not opt.noautoanchor:
                check_anchors(dataset, model=model, thr=hyp['anchor_t'], imgsz=imgsz)
            model.half().float()  # pre-reduce anchor precision

        callbacks.run('on_pretrain_routine_end')

    # DDP mode
    if cuda and RANK != -1:
        if train_yolo!='nothing':
            model = DDP(model, device_ids=[LOCAL_RANK], output_device=LOCAL_RANK)
        if not freeze_autoenc:
            autoencoder = DDP(autoencoder, device_ids=[LOCAL_RANK], output_device=LOCAL_RANK)
        rec_model = DDP(rec_model, device_ids=[LOCAL_RANK], output_device=LOCAL_RANK)

    # Model parameters
    nl = de_parallel(model).model[-1].nl  # number of detection layers (to scale hyps)
    hyp['box'] *= 3 / nl  # scale to layers
    hyp['cls'] *= nc / 80 * 3 / nl  # scale to classes and layers
    hyp['obj'] *= (imgsz / 640) ** 2 * 3 / nl  # scale to image size and layers
    hyp['label_smoothing'] = opt.label_smoothing
    model.nc = nc  # attach number of classes to model
    model.hyp = hyp  # attach hyperparameters to model
    model.class_weights = labels_to_class_weights(dataset.labels, nc).to(device) * nc  # attach class weights
    model.names = names

    # Start training
    t0 = time.time()
    nw = max(round(hyp['warmup_epochs'] * nb), 100)  # number of warmup iterations, max(3 epochs, 100 iterations)
    last_opt_step_obj = -1
    maps = np.zeros(nc)  # mAP per class
    results = (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0)  # P, R, mAP@.5, mAP@.5-.95, val_loss(box, obj, cls),  val_loss(compressibility), val_loss(mae, mse, psnr, grad)
    scheduler.last_epoch = start_epoch - 1  # do not move
    rec_scheduler.last_epoch = start_epoch - 1  # do not move
    scaler = amp.GradScaler(enabled=cuda)
    stopper = EarlyStopping(patience=opt.patience)
    compute_loss = ComputeLoss(model)  # init loss class
    compute_rec_loss = ComputeRecLoss(MAX=1.0, w_grad=opt.w_grad)  # init loss class
    mloss = torch.zeros(9, device=device)  # mean losses
    compressibility_loss = CompressibilityLoss(device, opt.w_compress)
    LOGGER.info(f'Image sizes {imgsz} train, {imgsz} val\n'
                f'Using {train_loader.num_workers} dataloader workers\n'
                f"Logging results to {colorstr('bold', save_dir)}\n"
                f'Starting training for {epochs} epochs...')
    for epoch in range(start_epoch, epochs):  # epoch ------------------------------------------------------------------
        # starting training the object detection model
        model.train()
        autoencoder.train()
        rec_model.train()

        # Update image weights (optional, single-GPU only)
        if opt.image_weights:
            cw = model.class_weights.cpu().numpy() * (1 - maps) ** 2 / nc  # class weights
            iw = labels_to_image_weights(dataset.labels, nc=nc, class_weights=cw)  # image weights
            dataset.indices = random.choices(range(dataset.n), weights=iw, k=dataset.n)  # rand weighted idx

        mloss = torch.zeros(9, device=device)  # mean losses
        if RANK != -1:
            train_loader.sampler.set_epoch(epoch)
        pbar = enumerate(train_loader)
        LOGGER.info(('%10s' * 13) % ('Epoch', 'gpu_mem', 'box', 'obj', 'cls', 'cmprss', 'mae', 'mse', 'psnr', 'grad', 'ms-ssim', 'labels', 'img_size'))
        if RANK in [-1, 0]:
            pbar = tqdm(pbar, total=nb)  # progress bar
        optimizer.zero_grad()
        rec_optimizer.zero_grad()
        for i, (imgs, targets, paths, _) in pbar:  # batch -------------------------------------------------------------
            ni = i + nb * epoch  # number integrated batches (since train start)
            imgs = imgs.to(device, non_blocking=True).float() / 255  # uint8 to float32, 0-255 to 0.0-1.0

            # Warmup
            if ni <= nw:
                xi = [0, nw]  # x interp
                accumulate = max(1, np.interp(ni, xi, [1, nbs / batch_size]).round())
                for j, x in enumerate(optimizer.param_groups):
                    # bias lr falls from 0.1 to lr0, all other lrs rise from 0.0 to lr0
                    x['lr'] = np.interp(ni, xi, [hyp['warmup_bias_lr'] if j == 2 else 0.0, x['initial_lr'] * lf(epoch)])
                    if 'momentum' in x:
                        x['momentum'] = np.interp(ni, xi, [hyp['warmup_momentum'], hyp['momentum']])
                for j, x in enumerate(rec_optimizer.param_groups):
                    # bias lr falls from 0.1 to lr0, all other lrs rise from 0.0 to lr0
                    x['lr'] = np.interp(ni, xi, [hyp['warmup_bias_lr'] if j == 2 else 0.0, x['initial_lr'] * lf(epoch)])
                    if 'momentum' in x:
                        x['momentum'] = np.interp(ni, xi, [hyp['warmup_momentum'], hyp['momentum']])

            # Multi-scale
            if opt.multi_scale:
                sz = random.randrange(imgsz * 0.5, imgsz * 1.5 + gs) // gs * gs  # size
                sf = sz / max(imgs.shape[2:])  # scale factor
                if sf != 1:
                    ns = [math.ceil(x * sf / gs) * gs for x in imgs.shape[2:]]  # new shape (stretched to gs-multiple)
                    imgs = nn.functional.interpolate(imgs, size=ns, mode='bilinear', align_corners=False)

            # First step of the adversarial training (train the RecNet)
            with amp.autocast(enabled=cuda):
                with torch.no_grad():
                    T = model(imgs, cut_model=1)  # first half of the model
                    T_bottleneck = autoencoder(T, task='enc')
                rec_imgs = rec_model(T_bottleneck)
                rec_loss, rec_loss_items, _ = compute_rec_loss(imgs, rec_imgs.float())
                if RANK != -1:
                    rec_loss *= WORLD_SIZE  # gradient averaged between devices in DDP mode
                if opt.quad:
                    rec_loss *= 4.

            # Backward
            scaler.scale(rec_loss).backward()

            if opt.clip_grad > 0:
                torch.nn.utils.clip_grad_value_(rec_model.parameters(), opt.clip_grad*10)
                torch.nn.utils.clip_grad_norm_(rec_model.parameters(), opt.clip_grad)

            # Optimize
            scaler.step(rec_optimizer)  # optimizer.step
            scaler.update()
            rec_optimizer.zero_grad()

            # Second step of the adversarial training (train the Autoencoder)
            with amp.autocast(enabled=cuda):
                T = model(imgs, cut_model=1)  # first half of the model
                T_bottleneck = autoencoder(T, task='enc')
                T_hat = autoencoder(T, task='dec', bottleneck=T_bottleneck)
                pred = model(None, cut_model=2, T=T_hat)  # second half of the model
                rec_imgs = rec_model(T_bottleneck)

                rec_loss, rec_loss_items, _ = compute_rec_loss(imgs, rec_imgs.float())
                loss_o, loss_items = compute_loss(pred, targets.to(device))  # loss scaled by batch_size
                loss_r, loss_items_r = compressibility_loss(T_bottleneck)
                loss_items = torch.cat((loss_items, loss_items_r, rec_loss_items))
                loss = loss_o + loss_r - opt.w_rec * rec_loss
                if RANK != -1:
                    loss *= WORLD_SIZE  # gradient averaged between devices in DDP mode
                if opt.quad:
                    loss *= 4.

            # Backward
            scaler.scale(loss).backward()

            if opt.clip_grad > 0:
                torch.nn.utils.clip_grad_value_(autoencoder.parameters(), opt.clip_grad*10)
                torch.nn.utils.clip_grad_norm_(autoencoder.parameters(), opt.clip_grad)

            # Optimize
            scaler.step(optimizer)  # optimizer.step
            scaler.update()
            optimizer.zero_grad()
            rec_optimizer.zero_grad()
            if ema:
                ema.update(model)

            # Log
            if RANK in [-1, 0]:
                mloss = (mloss * i + loss_items) / (i + 1)  # update mean losses
                mem = f'{torch.cuda.memory_reserved() / 1E9 if torch.cuda.is_available() else 0:.3g}G'  # (GB)
                pbar.set_description(('%10s' * 2 + '%10.4g' * 11) % (
                    f'{epoch}/{epochs - 1}', mem, *mloss, targets.shape[0], imgs.shape[-1]))
                callbacks.run('on_train_batch_end', ni, model, imgs, targets, paths, plots, opt.sync_bn)
            # end batch ------------------------------------------------------------------------------------------------

        # Scheduler
        lr = [x['lr'] for x in optimizer.param_groups]  # for loggers
        scheduler.step()
        rec_scheduler.step()

        # ------- validation -------- #
        rec_picture = rec_dir / Path(f'{epoch:03}.jpeg')
        if RANK in [-1, 0]:
            del T, T_bottleneck, T_hat, imgs, pred, rec_imgs, 
            torch.cuda.empty_cache()
            # mAP
            callbacks.run('on_train_epoch_end', epoch=epoch)
            ema.update_attr(model, include=['yaml', 'nc', 'hyp', 'names', 'stride', 'class_weights'])
            final_epoch = (epoch + 1 == epochs) or stopper.possible_stop
            if not noval or final_epoch:  # Calculate mAP
                results, maps, _ = val.run(data_dict,
                                        batch_size=batch_size // WORLD_SIZE * 2,
                                        imgsz=imgsz,
                                        model=ema.ema,
                                        single_cls=single_cls,
                                        dataloader=val_loader,
                                        save_dir=save_dir,
                                        plots=False,
                                        callbacks=callbacks,
                                        compute_loss=compute_loss,
                                        compressibility_loss=compressibility_loss,
                                        compute_rec_loss=compute_rec_loss,
                                        autoencoder=deepcopy(autoencoder),
                                        rec_model=deepcopy(rec_model))
                
                reconstruct.run(model=ema.ema,
                                save_dir=save_dir,
                                autoencoder=deepcopy(autoencoder),
                                rec_model=deepcopy(rec_model),
                                sample_img=opt.sample_img,
                                store_img = rec_picture)
     
            log_vals = list(mloss[-5:]) + list(results)[-5:]
            callbacks.run('on_train_epoch_end_adversary', log_vals, epoch, rec_picture)
        
            # Update best mAP
            rec_fitness = -np.array(results)[-2]
            fi = fitness_adv(np.array(results).reshape(1, -1), rec_fitness)  # weighted combination of [P, R, mAP@.5, mAP@.5-.95, psnr]
            if fi > best_fitness and epoch > 2:
                best_fitness = fi
                best_epoch = epoch
            log_vals = list(mloss[:4]) + list(results)[:-5] + lr
            callbacks.run('on_fit_epoch_end', log_vals, epoch, best_fitness, fi)

            # Save model
            if (not nosave) or (final_epoch and not evolve):  # if save
                ckpt = {'epoch': epoch,
                        'best_fitness': best_fitness,
                        'model': deepcopy(de_parallel(model)).half(),
                        'ema': deepcopy(ema.ema).half(),
                        'updates': ema.updates,
                        'optimizer': optimizer.state_dict(),
                        'wandb_id': loggers.wandb.wandb_run.id if loggers.wandb else None,
                        'date': datetime.now().isoformat(),
                        'autoencoder': deepcopy(de_parallel(autoencoder)).half(),
                        'rec_model': deepcopy(de_parallel(rec_model)).half()}

                # Save last, best and delete
                torch.save(ckpt, last)
                if best_fitness == fi:
                    torch.save(ckpt, best)
                if (epoch > 0) and (opt.save_period > 0) and (epoch % opt.save_period == 0):
                    torch.save(ckpt, w / f'epoch{epoch}.pt')
                del ckpt
                callbacks.run('on_model_save', last, epoch, final_epoch, best_fitness, fi)

            # Stop Single-GPU
            if RANK == -1 and stopper(epoch=epoch, fitness=fi):
                break
        # end epoch -------------------------------------------------------------
    LOGGER.info(f'\nThe best fitness is {best_fitness[0]:.4g} associated with epoch {best_epoch}')    
    # end training -----------------------------------------------------------------------------------------------------

    
    if RANK in [-1, 0]:
        del model, autoencoder, rec_model
        del optimizer, rec_optimizer
        torch.cuda.empty_cache()
        LOGGER.info(f'\n{epoch - start_epoch + 1} epochs completed in {(time.time() - t0) / 3600:.3f} hours.')
        rec_picture = rec_dir / Path(f'{best_epoch:03}.jpeg')
        for f in last, best:
            if f.exists():
                strip_optimizer(f)  # strip optimizers
                if f is best:

                    ckpt = torch.load(f, map_location=device)
                    best_autoencoder = AutoEncoder(autoenc_chs).to(device)
                    best_autoencoder.load_state_dict(ckpt['autoencoder'].state_dict())
                    best_rec_model = Decoder_Rec(cin=ckpt['rec_model'].cin, cout=ckpt['rec_model'].cout, first_chs=autoenc_chs).to(device)
                    best_rec_model.load_state_dict(ckpt['rec_model'].state_dict())
                    model=attempt_load(f, device).half()

                    LOGGER.info(f'\nValidating {f}...')
                    results, _, _ = val.run(data_dict,
                                            batch_size=batch_size // WORLD_SIZE * 2,
                                            imgsz=imgsz,
                                            model=model,
                                            iou_thres=0.65 if is_coco else 0.60,  # best pycocotools results at 0.65
                                            single_cls=single_cls,
                                            dataloader=val_loader,
                                            save_dir=save_dir,
                                            save_json=is_coco,
                                            verbose=True,
                                            plots=True,
                                            callbacks=callbacks,
                                            compute_loss=compute_loss,      # val best model with plots
                                            compute_rec_loss=compute_rec_loss,
                                            autoencoder=best_autoencoder,
                                            rec_model=best_rec_model)
                    
                    reconstruct.run(model=model,
                                    save_dir=save_dir,
                                    autoencoder=best_autoencoder,
                                    rec_model=best_rec_model,
                                    sample_img=opt.sample_img,
                                    store_img = rec_picture)
                    
                    log_vals = list(mloss[-5:]) + list(results)[-5:]
                    callbacks.run('on_train_epoch_end_adversary', log_vals, best_epoch, rec_picture)
                    log_vals = list(mloss[:4]) + list(results)[:-5] + lr
                    callbacks.run('on_fit_epoch_end', log_vals, epoch, best_fitness, fi)

        callbacks.run('on_train_end', last, best, plots, epoch, results)
        LOGGER.info(f"Results saved to {colorstr('bold', save_dir)}")

    torch.cuda.empty_cache()
    return results


def parse_opt(known=False):
    parser = argparse.ArgumentParser()
    parser.add_argument('--weights', type=str, default=ROOT / 'yolov5m.pt', help='initial weights path')
    parser.add_argument('--cfg', type=str, default='', help='model.yaml path')
    parser.add_argument('--data', type=str, default=ROOT / 'data/coco128.yaml', help='dataset.yaml path')
    parser.add_argument('--hyp', type=str, default=ROOT / 'data/hyps/hyp.scratch-high.yaml', help='hyperparameters path')
    parser.add_argument('--epochs', type=int, default=40)
    parser.add_argument('--batch-size', type=int, default=16, help='total batch size for all GPUs, -1 for autobatch')
    parser.add_argument('--imgsz', '--img', '--img-size', type=int, default=512, help='train, val image size (pixels)')
    parser.add_argument('--rect', action='store_true', help='rectangular training')
    parser.add_argument('--resume', nargs='?', const=True, default=False, help='resume most recent training')
    parser.add_argument('--nosave', action='store_true', help='only save final checkpoint')
    parser.add_argument('--noval', action='store_true', help='only validate final epoch')
    parser.add_argument('--noautoanchor', action='store_true', help='disable autoanchor check')
    parser.add_argument('--evolve', type=int, nargs='?', const=300, help='evolve hyperparameters for x generations')
    parser.add_argument('--bucket', type=str, default='', help='gsutil bucket')
    parser.add_argument('--cache', type=str, nargs='?', const='ram', help='--cache images in "ram" (default) or "disk"')
    parser.add_argument('--image-weights', action='store_true', help='use weighted image selection for training')
    parser.add_argument('--device', default='', help='cuda device, i.e. 0 or 0,1,2,3 or cpu')
    parser.add_argument('--multi-scale', action='store_true', help='vary img-size +/- 50%%')
    parser.add_argument('--single-cls', action='store_true', help='train multi-class data as single-class')
    parser.add_argument('--adam', action='store_true', help='use torch.optim.Adam() optimizer')
    parser.add_argument('--sync-bn', action='store_true', help='use SyncBatchNorm, only available in DDP mode')
    parser.add_argument('--workers', type=int, default=8, help='maximum number of dataloader workers')
    parser.add_argument('--project', default=ROOT / 'runs/train_adv', help='save to project/name')
    parser.add_argument('--name', default='exp', help='save to project/name')
    parser.add_argument('--exist-ok', action='store_true', help='existing project/name ok, do not increment')
    parser.add_argument('--quad', action='store_true', help='quad dataloader')
    parser.add_argument('--linear-lr', action='store_true', help='linear LR')
    parser.add_argument('--label-smoothing', type=float, default=0.0, help='Label smoothing epsilon')
    parser.add_argument('--patience', type=int, default=100, help='EarlyStopping patience (epochs without improvement)')
    parser.add_argument('--freeze', type=int, default=None, help='Number of layers to freeze. backbone=10, all=24')
    parser.add_argument('--save-period', type=int, default=-1, help='Save checkpoint every x epochs (disabled if < 1)')
    parser.add_argument('--local_rank', type=int, default=-1, help='DDP parameter, do not modify')
    # Weights & Biases arguments
    parser.add_argument('--entity', default=None, help='W&B: Entity')
    parser.add_argument('--upload_dataset', action='store_true', help='W&B: Upload dataset as artifact table')
    parser.add_argument('--bbox_interval', type=int, default=-1, help='W&B: Set bounding-box image logging interval')
    parser.add_argument('--artifact_alias', type=str, default='latest', help='W&B: Version of dataset artifact to use')
    # Supplemental arguments
    parser.add_argument('--autoenc-chs',  type=int, nargs='*', default=[], help='number of channels in autoencoder suggested: [192, 128, 64]')
    parser.add_argument('--train-yolo', type=str, default='nothing', help='which part of the yolo gets trained: backend, all, nothing')
    parser.add_argument('--freeze-autoenc', action='store_true', help='determins autoencoder weights to be freezed')
    parser.add_argument('--rec-only', action='store_true', help='only train the reconstruction netweork')
    parser.add_argument('--cut-layer', type=int, default=4, help='the index of the cutting layer (AFTER this layer, the model will be split) (ONLY 4 is supported for now)')
    # Some of the hyperparameters
    parser.add_argument('--lr0', type=float, default=0.01, help='initial learning rate')
    parser.add_argument('--lrf', type=float, default=0.002, help='final OneCycleLR learning rate (lr0 * lrf)')
    parser.add_argument('--warmup-bias-lr', type=float, default=0.1, help='warmup initial bias lr')
    parser.add_argument('--weight-decay', type=float, default=0.0005, help='optimizer weight decay 5e-4')
    parser.add_argument('--w-compress', type=float, default=0, help='The weight of the compressibility loss')
    parser.add_argument('--w-rec', type=float, default=0, help='The weight of the reconstruction loss')
    parser.add_argument('--w-grad', type=float, default=5, help='The weight of the image gradient loss in reconstruction loss')
    parser.add_argument('--clip-grad', type=float, default=-1, help='The max norm of the clipped gradient')
    parser.add_argument('--sample-img', type=str, default=None, help='The sample image to be reconstructed and kept track during training')
    

    opt = parser.parse_known_args()[0] if known else parser.parse_args()
    return opt


def main(opt, callbacks=Callbacks()):
    # Checks
    if RANK in [-1, 0]:
        print_args(FILE.stem, opt)
        check_git_status()
        check_requirements(exclude=['thop'])

    # Resume
    if opt.resume and not check_wandb_resume(opt) and not opt.evolve:  # resume an interrupted run
        ckpt = opt.resume if isinstance(opt.resume, str) else get_latest_run()  # specified or most recent path
        assert os.path.isfile(ckpt), 'ERROR: --resume checkpoint does not exist'
        with open(Path(ckpt).parent.parent / 'opt.yaml', errors='ignore') as f:
            opt = argparse.Namespace(**yaml.safe_load(f))  # replace
        opt.cfg, opt.weights, opt.resume = '', ckpt, True  # reinstate
        LOGGER.info(f'Resuming training from {ckpt}')
    else:
        opt.data, opt.cfg, opt.hyp, opt.weights, opt.project = \
            check_file(opt.data), check_yaml(opt.cfg), check_yaml(opt.hyp), str(opt.weights), str(opt.project)  # checks
        assert len(opt.cfg) or len(opt.weights), 'either --cfg or --weights must be specified'
        if opt.evolve:
            opt.project = str(ROOT / 'runs/evolve')
            opt.exist_ok, opt.resume = opt.resume, False  # pass resume to exist_ok and disable resume
        opt.save_dir = str(increment_path(Path(opt.project) / opt.name, exist_ok=opt.exist_ok))

    # DDP mode
    device = select_device(opt.device, batch_size=opt.batch_size)
    if LOCAL_RANK != -1:
        assert torch.cuda.device_count() > LOCAL_RANK, 'insufficient CUDA devices for DDP command'
        assert opt.batch_size % WORLD_SIZE == 0, '--batch-size must be multiple of CUDA device count'
        assert not opt.image_weights, '--image-weights argument is not compatible with DDP training'
        assert not opt.evolve, '--evolve argument is not compatible with DDP training'
        torch.cuda.set_device(LOCAL_RANK)
        device = torch.device('cuda', LOCAL_RANK)
        dist.init_process_group(backend="nccl" if dist.is_nccl_available() else "gloo")

    # Train
    if not opt.evolve:
        train(opt.hyp, opt, device, callbacks)
        if WORLD_SIZE > 1 and RANK == 0:
            LOGGER.info('Destroying process group... ')
            dist.destroy_process_group()

    # Evolve hyperparameters (optional)
    else:
        # Hyperparameter evolution metadata (mutation scale 0-1, lower_limit, upper_limit)
        meta = {'lr0': (1, 1e-5, 1e-1),  # initial learning rate (SGD=1E-2, Adam=1E-3)
                'lrf': (1, 0.01, 1.0),  # final OneCycleLR learning rate (lr0 * lrf)
                'momentum': (0.3, 0.6, 0.98),  # SGD momentum/Adam beta1
                'weight_decay': (1, 0.0, 0.001),  # optimizer weight decay
                'warmup_epochs': (1, 0.0, 5.0),  # warmup epochs (fractions ok)
                'warmup_momentum': (1, 0.0, 0.95),  # warmup initial momentum
                'warmup_bias_lr': (1, 0.0, 0.2),  # warmup initial bias lr
                'box': (1, 0.02, 0.2),  # box loss gain
                'cls': (1, 0.2, 4.0),  # cls loss gain
                'cls_pw': (1, 0.5, 2.0),  # cls BCELoss positive_weight
                'obj': (1, 0.2, 4.0),  # obj loss gain (scale with pixels)
                'obj_pw': (1, 0.5, 2.0),  # obj BCELoss positive_weight
                'iou_t': (0, 0.1, 0.7),  # IoU training threshold
                'anchor_t': (1, 2.0, 8.0),  # anchor-multiple threshold
                'anchors': (2, 2.0, 10.0),  # anchors per output grid (0 to ignore)
                'fl_gamma': (0, 0.0, 2.0),  # focal loss gamma (efficientDet default gamma=1.5)
                'hsv_h': (1, 0.0, 0.1),  # image HSV-Hue augmentation (fraction)
                'hsv_s': (1, 0.0, 0.9),  # image HSV-Saturation augmentation (fraction)
                'hsv_v': (1, 0.0, 0.9),  # image HSV-Value augmentation (fraction)
                'degrees': (1, 0.0, 45.0),  # image rotation (+/- deg)
                'translate': (1, 0.0, 0.9),  # image translation (+/- fraction)
                'scale': (1, 0.0, 0.9),  # image scale (+/- gain)
                'shear': (1, 0.0, 10.0),  # image shear (+/- deg)
                'perspective': (0, 0.0, 0.001),  # image perspective (+/- fraction), range 0-0.001
                'flipud': (1, 0.0, 1.0),  # image flip up-down (probability)
                'fliplr': (0, 0.0, 1.0),  # image flip left-right (probability)
                'mosaic': (1, 0.0, 1.0),  # image mixup (probability)
                'mixup': (1, 0.0, 1.0),  # image mixup (probability)
                'copy_paste': (1, 0.0, 1.0)}  # segment copy-paste (probability)

        with open(opt.hyp, errors='ignore') as f:
            hyp = yaml.safe_load(f)  # load hyps dict
            if 'anchors' not in hyp:  # anchors commented in hyp.yaml
                hyp['anchors'] = 3
        opt.noval, opt.nosave, save_dir = True, True, Path(opt.save_dir)  # only val/save final epoch
        # ei = [isinstance(x, (int, float)) for x in hyp.values()]  # evolvable indices
        evolve_yaml, evolve_csv = save_dir / 'hyp_evolve.yaml', save_dir / 'evolve.csv'
        if opt.bucket:
            os.system(f'gsutil cp gs://{opt.bucket}/evolve.csv {save_dir}')  # download evolve.csv if exists

        for _ in range(opt.evolve):  # generations to evolve
            if evolve_csv.exists():  # if evolve.csv exists: select best hyps and mutate
                # Select parent(s)
                parent = 'single'  # parent selection method: 'single' or 'weighted'
                x = np.loadtxt(evolve_csv, ndmin=2, delimiter=',', skiprows=1)
                n = min(5, len(x))  # number of previous results to consider
                x = x[np.argsort(-fitness(x))][:n]  # top n mutations
                w = fitness(x) - fitness(x).min() + 1E-6  # weights (sum > 0)
                if parent == 'single' or len(x) == 1:
                    # x = x[random.randint(0, n - 1)]  # random selection
                    x = x[random.choices(range(n), weights=w)[0]]  # weighted selection
                elif parent == 'weighted':
                    x = (x * w.reshape(n, 1)).sum(0) / w.sum()  # weighted combination

                # Mutate
                mp, s = 0.8, 0.2  # mutation probability, sigma
                npr = np.random
                npr.seed(int(time.time()))
                g = np.array([meta[k][0] for k in hyp.keys()])  # gains 0-1
                ng = len(meta)
                v = np.ones(ng)
                while all(v == 1):  # mutate until a change occurs (prevent duplicates)
                    v = (g * (npr.random(ng) < mp) * npr.randn(ng) * npr.random() * s + 1).clip(0.3, 3.0)
                for i, k in enumerate(hyp.keys()):  # plt.hist(v.ravel(), 300)
                    hyp[k] = float(x[i + 7] * v[i])  # mutate

            # Constrain to limits
            for k, v in meta.items():
                hyp[k] = max(hyp[k], v[1])  # lower limit
                hyp[k] = min(hyp[k], v[2])  # upper limit
                hyp[k] = round(hyp[k], 5)  # significant digits

            # Train mutation
            results = train(hyp.copy(), opt, device, callbacks)

            # Write mutation results
            print_mutation(results, hyp.copy(), save_dir, opt.bucket)

        # Plot results
        plot_evolve(evolve_csv)
        LOGGER.info(f'Hyperparameter evolution finished\n'
                    f"Results saved to {colorstr('bold', save_dir)}\n"
                    f'Use best hyperparameters example: $ python train.py --hyp {evolve_yaml}')


def run(**kwargs):
    # Usage: import train; train.run(data='coco128.yaml', imgsz=320, weights='yolov5m.pt')
    opt = parse_opt(True)
    for k, v in kwargs.items():
        setattr(opt, k, v)
    main(opt)


if __name__ == "__main__":
    opt = parse_opt()
    main(opt)