utils.py

# Copyright (c) 2015-present, Facebook, Inc.
# All rights reserved.
"""
Misc functions, including distributed helpers.
Mostly copy-paste from torchvision references.
"""
import io
import os
import time
from collections import defaultdict, deque
import datetime
from statistics import mean, stdev

import numpy as np
from matplotlib import pyplot as plt
from matplotlib.lines import Line2D

import torch
import torch.distributed as dist
from torchvision.utils import save_image
from thop import profile
from torchprofile import profile_macs
from ptflops import get_model_complexity_info


class SmoothedValue(object):
    """Track a series of values and provide access to smoothed values over a
    window or the global series average.
    """

    def __init__(self, window_size=20, fmt=None):
        if fmt is None:
            fmt = "{median:.4f} ({global_avg:.4f})"
        self.deque = deque(maxlen=window_size)
        self.total = 0.0
        self.count = 0
        self.fmt = fmt

    def update(self, value, n=1):
        self.deque.append(value)
        self.count += n
        self.total += value * n

    def synchronize_between_processes(self):
        """
        Warning: does not synchronize the deque!
        """
        if not is_dist_avail_and_initialized():
            return
        t = torch.tensor([self.count, self.total], dtype=torch.float64, device='cuda')
        dist.barrier()
        dist.all_reduce(t)
        t = t.tolist()
        self.count = int(t[0])
        self.total = t[1]

    @property
    def median(self):
        d = torch.tensor(list(self.deque))
        return d.median().item()

    @property
    def avg(self):
        d = torch.tensor(list(self.deque), dtype=torch.float32)
        return d.mean().item()

    @property
    def global_avg(self):
        return self.total / self.count

    @property
    def max(self):
        return max(self.deque)

    @property
    def value(self):
        return self.deque[-1]

    def __str__(self):
        return self.fmt.format(
            median=self.median,
            avg=self.avg,
            global_avg=self.global_avg,
            max=self.max,
            value=self.value)


class MetricLogger(object):
    def __init__(self, delimiter="\t"):
        self.meters = defaultdict(SmoothedValue)
        self.delimiter = delimiter

    def update(self, **kwargs):
        for k, v in kwargs.items():
            if isinstance(v, torch.Tensor):
                v = v.item()
            assert isinstance(v, (float, int))
            self.meters[k].update(v)

    def __getattr__(self, attr):
        if attr in self.meters:
            return self.meters[attr]
        if attr in self.__dict__:
            return self.__dict__[attr]
        raise AttributeError("'{}' object has no attribute '{}'".format(
            type(self).__name__, attr))

    def __str__(self):
        loss_str = []
        for name, meter in self.meters.items():
            loss_str.append(
                "{}: {}".format(name, str(meter))
            )
        return self.delimiter.join(loss_str)

    def synchronize_between_processes(self):
        for meter in self.meters.values():
            meter.synchronize_between_processes()

    def add_meter(self, name, meter):
        self.meters[name] = meter

    def log_every(self, iterable, print_freq, header=None):
        i = 0
        if not header:
            header = ''
        start_time = time.time()
        end = time.time()
        iter_time = SmoothedValue(fmt='{avg:.4f}')
        data_time = SmoothedValue(fmt='{avg:.4f}')
        space_fmt = ':' + str(len(str(len(iterable)))) + 'd'
        log_msg = [
            header,
            '[{0' + space_fmt + '}/{1}]',
            'eta: {eta}',
            '{meters}',
            'time: {time}',
            'data: {data}'
        ]
        if torch.cuda.is_available():
            log_msg.append('max mem: {memory:.0f}')
        log_msg = self.delimiter.join(log_msg)
        MB = 1024.0 * 1024.0
        for obj in iterable:
            data_time.update(time.time() - end)
            yield obj
            iter_time.update(time.time() - end)
            if i % print_freq == 0 or i == len(iterable) - 1:
                eta_seconds = iter_time.global_avg * (len(iterable) - i)
                eta_string = str(datetime.timedelta(seconds=int(eta_seconds)))
                if torch.cuda.is_available():
                    print(log_msg.format(
                        i, len(iterable), eta=eta_string,
                        meters=str(self),
                        time=str(iter_time), data=str(data_time),
                        memory=torch.cuda.max_memory_allocated() / MB))
                else:
                    print(log_msg.format(
                        i, len(iterable), eta=eta_string,
                        meters=str(self),
                        time=str(iter_time), data=str(data_time)))
            i += 1
            end = time.time()
        total_time = time.time() - start_time
        total_time_str = str(datetime.timedelta(seconds=int(total_time)))
        print('{} Total time: {} ({:.4f} s / it)'.format(
            header, total_time_str, total_time / len(iterable)))


def _load_checkpoint_for_ema(model_ema, checkpoint):
    """
    Workaround for ModelEma._load_checkpoint to accept an already-loaded object
    """
    mem_file = io.BytesIO()
    torch.save({'state_dict_ema':checkpoint}, mem_file)
    mem_file.seek(0)
    model_ema._load_checkpoint(mem_file)


def setup_for_distributed(is_master):
    """
    This function disables printing when not in master process
    """
    import builtins as __builtin__
    builtin_print = __builtin__.print

    def print(*args, **kwargs):
        force = kwargs.pop('force', False)
        if is_master or force:
            builtin_print(*args, **kwargs)

    __builtin__.print = print


def is_dist_avail_and_initialized():
    if not dist.is_available():
        return False
    if not dist.is_initialized():
        return False
    return True


def get_world_size():
    if not is_dist_avail_and_initialized():
        return 1
    return dist.get_world_size()


def get_rank():
    if not is_dist_avail_and_initialized():
        return 0
    return dist.get_rank()


def is_main_process():
    return get_rank() == 0


def save_on_master(*args, **kwargs):
    if is_main_process():
        torch.save(*args, **kwargs)


def init_distributed_mode(args):
    if 'RANK' in os.environ and 'WORLD_SIZE' in os.environ:
        args.rank = int(os.environ["RANK"])
        args.world_size = int(os.environ['WORLD_SIZE'])
        args.gpu = int(os.environ['LOCAL_RANK'])
    #elif 'SLURM_PROCID' in os.environ:
    #    args.rank = int(os.environ['SLURM_PROCID'])
    #    args.gpu = args.rank % torch.cuda.device_count()
    else:
        print('Not using distributed mode')
        args.distributed = False
        return

    args.distributed = True

    torch.cuda.set_device(args.gpu)
    args.dist_backend = 'nccl'
    print('| distributed init (rank {}): {}'.format(
        args.rank, args.dist_url), flush=True)
    torch.distributed.init_process_group(backend=args.dist_backend, init_method=args.dist_url,
                                         world_size=args.world_size, rank=args.rank)
    torch.distributed.barrier()
    setup_for_distributed(args.rank == 0)



def average_precision(output, target):
    epsilon = 1e-8

    # sort examples
    indices = output.argsort()[::-1]
    # Computes prec@i
    total_count_ = np.cumsum(np.ones((len(output), 1)))

    target_ = target[indices]
    ind = target_ == 1
    pos_count_ = np.cumsum(ind)
    total = pos_count_[-1]
    pos_count_[np.logical_not(ind)] = 0
    pp = pos_count_ / total_count_
    precision_at_i_ = np.sum(pp)
    precision_at_i = precision_at_i_ / (total + epsilon)

    return precision_at_i


def mAP(targs, preds):
    """Returns the model's average precision for each class
    Return:
        ap (FloatTensor): 1xK tensor, with avg precision for each class k
    """

    if np.size(preds) == 0:
        return 0
    ap = np.zeros((preds.shape[1]))
    # compute average precision for each class
    for k in range(preds.shape[1]):
        # sort scores
        scores = preds[:, k]
        targets = targs[:, k]
        # compute average precision
        ap[k] = average_precision(scores, targets)
    return 100 * ap.mean()


def inverse_normalize(tensor):
    mean = (0.485, 0.456, 0.406)
    std = (0.229, 0.224, 0.225)

    mean = torch.as_tensor(mean, dtype=tensor.dtype, device=tensor.device)
    std = torch.as_tensor(std, dtype=tensor.dtype, device=tensor.device)
    if mean.ndim == 1:
        mean = mean.view(-1, 1, 1)
    if std.ndim == 1:
        std = std.view(-1, 1, 1)
    tensor.mul_(std).add_(mean)
    return tensor


def save_images(output, output_dir, total_step):
    if len(output) == 3:
        _, _, augs = output
    elif len(output) == 4:
        _, _, _, augs = output
    else:
        return None

    augs = augs.reshape(-1, 3, augs.shape[-1], augs.shape[-1])
    augs = inverse_normalize(augs.data)

    fp = os.path.join(output_dir, f'augs_{total_step}.png')
    save_image(augs, fp, nrow=int(augs.shape[0] ** 0.5))

    return augs

def get_gradients(named_parameters):
    avg_grads = []
    max_grads= []
    layers = []
    for n, p in named_parameters:
        if(p.requires_grad) and ("bias" not in n):
            layers.append(n)
            avg_grads.append(p.grad.abs().mean().item())
            max_grads.append(p.grad.abs().max().item())

    max_grad = max(max_grads)
    mean_max_grad = mean(max_grads)
    std_max_grad = stdev(max_grads)
    
    mean_avg_grad = mean(avg_grads)
    std_avg_grad = stdev(avg_grads)

    gradients = {
        'max_grad_all': max_grad,
        'mean_max_grad_layer': mean_max_grad,
        'std_max_grad_layer': std_max_grad,
        'mean_avg_grad_layer': mean_avg_grad,
        'std_avg_grad_layer': std_avg_grad,
    }

    return gradients


# Based on https://discuss.pytorch.org/t/check-gradient-flow-in-network/15063/10
# @torch.no_grad()
def plot_grad_flow(named_parameters):
    '''Plots the gradients flowing through different layers in the net during training.
    Can be used for checking for possible gradient vanishing / exploding problems.
    
    Usage: Plug this function in Trainer class after loss.backwards() as 
    "plot_grad_flow(self.model.named_parameters())" to visualize the gradient flow'''
    avg_grads = []
    max_grads= []
    layers = []
    for n, p in named_parameters:
        if(p.requires_grad) and ("bias" not in n):
            layers.append(n)
            avg_grads.append(p.grad.abs().mean().item())
            max_grads.append(p.grad.abs().max().item())

    fig, ax = plt.subplot(figsize=(6, 4))

    ax.bar(np.arange(len(max_grads)), max_grads, alpha=0.1, lw=1, color="c")
    ax.bar(np.arange(len(max_grads)), avg_grads, alpha=0.1, lw=1, color="b")

    ax.hlines(0, 0, len(avg_grads)+1, lw=2, color="k" )
    ax.xticks(range(0,len(avg_grads), 1), layers, rotation="vertical")
    ax.xlim(left=0, right=len(avg_grads))
    ax.ylim(bottom = -0.001, top=0.02) # zoom in on the lower gradient regions
    ax.xlabel("Layers")
    ax.ylabel("average gradient")
    ax.title("Gradient flow")
    ax.grid(True)
    ax.legend([Line2D([0], [0], color="c", lw=4),
                Line2D([0], [0], color="b", lw=4),
                Line2D([0], [0], color="k", lw=4)], ['max-gradient', 'mean-gradient', 'zero-gradient'])

    return fig


def count_flops(model, image_size=224, device='cuda', debugging=False, profiler='torchprofile'):
    # https://github.com/sovrasov/flops-counter.pytorch/issues/16
    # https://github.com/Lyken17/pytorch-OpCounter
    # https://github.com/sovrasov/flops-counter.pytorch
    # https://github.com/zhijian-liu/torchprofile
    model.eval()

    with torch.no_grad():
        x = torch.rand(1, 3, image_size, image_size).to(device)
        if profiler == 'torchprofile':
            flops = profile_macs(model, x)
        elif profiler == 'ptflops':
            flops, _ = get_model_complexity_info(
                model,  x.shape[1:], as_strings=False,
                print_per_layer_stat=debugging, verbose=debugging)
        else:
            flops, _ = profile(model, inputs=(x, ))

    return flops