import torch import torch.nn as nn import torch.nn.functional as F from torchlibrosa.stft import Spectrogram, LogmelFilterBank from torchlibrosa.augmentation import SpecAugmentation from audio_infer.pytorch.pytorch_utils import do_mixup, interpolate, pad_framewise_output import os import sys import math import numpy as np import torch import torch.nn as nn import torch.nn.functional as F from torch.nn.parameter import Parameter from torchlibrosa.stft import Spectrogram, LogmelFilterBank from torchlibrosa.augmentation import SpecAugmentation from audio_infer.pytorch.pytorch_utils import do_mixup import torch.utils.checkpoint as checkpoint from timm.models.layers import DropPath, to_2tuple, trunc_normal_ import warnings from functools import partial #from mmdet.models.builder import BACKBONES from mmdet.utils import get_root_logger from mmcv.runner import load_checkpoint os.environ['TORCH_HOME'] = '../pretrained_models' from copy import deepcopy from timm.models.helpers import load_pretrained from torch.cuda.amp import autocast from collections import OrderedDict import io import re from mmcv.runner import _load_checkpoint, load_state_dict import mmcv.runner import copy import random from einops import rearrange from einops.layers.torch import Rearrange, Reduce from torch import nn, einsum def load_checkpoint(model, filename, map_location=None, strict=False, logger=None, revise_keys=[(r'^module\.', '')]): """Load checkpoint from a file or URI. Args: model (Module): Module to load checkpoint. filename (str): Accept local filepath, URL, ``torchvision://xxx``, ``open-mmlab://xxx``. Please refer to ``docs/model_zoo.md`` for details. map_location (str): Same as :func:`torch.load`. strict (bool): Whether to allow different params for the model and checkpoint. logger (:mod:`logging.Logger` or None): The logger for error message. revise_keys (list): A list of customized keywords to modify the state_dict in checkpoint. Each item is a (pattern, replacement) pair of the regular expression operations. Default: strip the prefix 'module.' by [(r'^module\\.', '')]. Returns: dict or OrderedDict: The loaded checkpoint. """ checkpoint = _load_checkpoint(filename, map_location, logger) new_proj = torch.nn.Conv2d(1, 64, kernel_size=(7, 7), stride=(4, 4), padding=(2, 2)) new_proj.weight = torch.nn.Parameter(torch.sum(checkpoint['patch_embed1.proj.weight'], dim=1).unsqueeze(1)) checkpoint['patch_embed1.proj.weight'] = new_proj.weight # OrderedDict is a subclass of dict if not isinstance(checkpoint, dict): raise RuntimeError( f'No state_dict found in checkpoint file {filename}') # get state_dict from checkpoint if 'state_dict' in checkpoint: state_dict = checkpoint['state_dict'] else: state_dict = checkpoint # strip prefix of state_dict metadata = getattr(state_dict, '_metadata', OrderedDict()) for p, r in revise_keys: state_dict = OrderedDict( {re.sub(p, r, k): v for k, v in state_dict.items()}) state_dict = OrderedDict({k.replace('backbone.',''):v for k,v in state_dict.items()}) # Keep metadata in state_dict state_dict._metadata = metadata # load state_dict load_state_dict(model, state_dict, strict, logger) return checkpoint def init_layer(layer): """Initialize a Linear or Convolutional layer. """ nn.init.xavier_uniform_(layer.weight) if hasattr(layer, 'bias'): if layer.bias is not None: layer.bias.data.fill_(0.) def init_bn(bn): """Initialize a Batchnorm layer. """ bn.bias.data.fill_(0.) bn.weight.data.fill_(1.) class TimeShift(nn.Module): def __init__(self, mean, std): super().__init__() self.mean = mean self.std = std def forward(self, x): if self.training: shift = torch.empty(1).normal_(self.mean, self.std).int().item() x = torch.roll(x, shift, dims=2) return x class LinearSoftPool(nn.Module): """LinearSoftPool Linear softmax, takes logits and returns a probability, near to the actual maximum value. Taken from the paper: A Comparison of Five Multiple Instance Learning Pooling Functions for Sound Event Detection with Weak Labeling https://arxiv.org/abs/1810.09050 """ def __init__(self, pooldim=1): super().__init__() self.pooldim = pooldim def forward(self, logits, time_decision): return (time_decision**2).sum(self.pooldim) / time_decision.sum( self.pooldim) class PVT(nn.Module): def __init__(self, sample_rate, window_size, hop_size, mel_bins, fmin, fmax, classes_num): super(PVT, self).__init__() window = 'hann' center = True pad_mode = 'reflect' ref = 1.0 amin = 1e-10 top_db = None # Spectrogram extractor self.spectrogram_extractor = Spectrogram(n_fft=window_size, hop_length=hop_size, win_length=window_size, window=window, center=center, pad_mode=pad_mode, freeze_parameters=True) # Logmel feature extractor self.logmel_extractor = LogmelFilterBank(sr=sample_rate, n_fft=window_size, n_mels=mel_bins, fmin=fmin, fmax=fmax, ref=ref, amin=amin, top_db=top_db, freeze_parameters=True) self.time_shift = TimeShift(0, 10) # Spec augmenter self.spec_augmenter = SpecAugmentation(time_drop_width=64, time_stripes_num=2, freq_drop_width=8, freq_stripes_num=2) self.bn0 = nn.BatchNorm2d(64) self.pvt_transformer = PyramidVisionTransformerV2(tdim=1001, fdim=64, patch_size=7, stride=4, in_chans=1, num_classes=classes_num, embed_dims=[64, 128, 320, 512], depths=[3, 4, 6, 3], num_heads=[1, 2, 5, 8], mlp_ratios=[8, 8, 4, 4], qkv_bias=True, qk_scale=None, drop_rate=0.0, drop_path_rate=0.1, sr_ratios=[8, 4, 2, 1], norm_layer=partial(nn.LayerNorm, eps=1e-6), num_stages=4, #pretrained='https://github.com/whai362/PVT/releases/download/v2/pvt_v2_b2.pth' ) #self.temp_pool = LinearSoftPool() self.avgpool = nn.AdaptiveAvgPool1d(1) self.fc_audioset = nn.Linear(512, classes_num, bias=True) self.init_weights() def init_weights(self): init_bn(self.bn0) init_layer(self.fc_audioset) def forward(self, input, mixup_lambda=None): """Input: (batch_size, times_steps, freq_bins)""" interpolate_ratio = 32 x = self.spectrogram_extractor(input) # (batch_size, 1, time_steps, freq_bins) x = self.logmel_extractor(x) # (batch_size, 1, time_steps, mel_bins) frames_num = x.shape[2] x = x.transpose(1, 3) x = self.bn0(x) x = x.transpose(1, 3) if self.training: x = self.time_shift(x) x = self.spec_augmenter(x) # Mixup on spectrogram if self.training and mixup_lambda is not None: x = do_mixup(x, mixup_lambda) #print(x.shape) #torch.Size([10, 1, 1001, 64]) x = self.pvt_transformer(x) #print(x.shape) #torch.Size([10, 800, 128]) x = torch.mean(x, dim=3) x = x.transpose(1, 2).contiguous() framewise_output = torch.sigmoid(self.fc_audioset(x)) #clipwise_output = torch.mean(framewise_output, dim=1) #clipwise_output = self.temp_pool(x, framewise_output).clamp(1e-7, 1.).squeeze(1) x = framewise_output.transpose(1, 2).contiguous() x = self.avgpool(x) clipwise_output = torch.flatten(x, 1) #print(framewise_output.shape) #torch.Size([10, 100, 17]) framewise_output = interpolate(framewise_output, interpolate_ratio) #framewise_output = framewise_output[:,:1000,:] #framewise_output = pad_framewise_output(framewise_output, frames_num) output_dict = {'framewise_output': framewise_output, 'clipwise_output': clipwise_output} return output_dict class PVT2(nn.Module): def __init__(self, sample_rate, window_size, hop_size, mel_bins, fmin, fmax, classes_num): super(PVT2, self).__init__() window = 'hann' center = True pad_mode = 'reflect' ref = 1.0 amin = 1e-10 top_db = None # Spectrogram extractor self.spectrogram_extractor = Spectrogram(n_fft=window_size, hop_length=hop_size, win_length=window_size, window=window, center=center, pad_mode=pad_mode, freeze_parameters=True) # Logmel feature extractor self.logmel_extractor = LogmelFilterBank(sr=sample_rate, n_fft=window_size, n_mels=mel_bins, fmin=fmin, fmax=fmax, ref=ref, amin=amin, top_db=top_db, freeze_parameters=True) self.time_shift = TimeShift(0, 10) # Spec augmenter self.spec_augmenter = SpecAugmentation(time_drop_width=64, time_stripes_num=2, freq_drop_width=8, freq_stripes_num=2) self.bn0 = nn.BatchNorm2d(64) self.pvt_transformer = PyramidVisionTransformerV2(tdim=1001, fdim=64, patch_size=7, stride=4, in_chans=1, num_classes=classes_num, embed_dims=[64, 128, 320, 512], depths=[3, 4, 6, 3], num_heads=[1, 2, 5, 8], mlp_ratios=[8, 8, 4, 4], qkv_bias=True, qk_scale=None, drop_rate=0.0, drop_path_rate=0.1, sr_ratios=[8, 4, 2, 1], norm_layer=partial(nn.LayerNorm, eps=1e-6), num_stages=4, pretrained='https://github.com/whai362/PVT/releases/download/v2/pvt_v2_b2.pth' ) #self.temp_pool = LinearSoftPool() self.fc_audioset = nn.Linear(512, classes_num, bias=True) self.init_weights() def init_weights(self): init_bn(self.bn0) init_layer(self.fc_audioset) def forward(self, input, mixup_lambda=None): """Input: (batch_size, times_steps, freq_bins)""" interpolate_ratio = 32 x = self.spectrogram_extractor(input) # (batch_size, 1, time_steps, freq_bins) x = self.logmel_extractor(x) # (batch_size, 1, time_steps, mel_bins) frames_num = x.shape[2] x = x.transpose(1, 3) x = self.bn0(x) x = x.transpose(1, 3) if self.training: #x = self.time_shift(x) x = self.spec_augmenter(x) # Mixup on spectrogram if self.training and mixup_lambda is not None: x = do_mixup(x, mixup_lambda) #print(x.shape) #torch.Size([10, 1, 1001, 64]) x = self.pvt_transformer(x) #print(x.shape) #torch.Size([10, 800, 128]) x = torch.mean(x, dim=3) x = x.transpose(1, 2).contiguous() framewise_output = torch.sigmoid(self.fc_audioset(x)) clipwise_output = torch.mean(framewise_output, dim=1) #clipwise_output = self.temp_pool(x, framewise_output).clamp(1e-7, 1.).squeeze(1) #print(framewise_output.shape) #torch.Size([10, 100, 17]) framewise_output = interpolate(framewise_output, interpolate_ratio) #framewise_output = framewise_output[:,:1000,:] #framewise_output = pad_framewise_output(framewise_output, frames_num) output_dict = {'framewise_output': framewise_output, 'clipwise_output': clipwise_output} return output_dict class PVT_2layer(nn.Module): def __init__(self, sample_rate, window_size, hop_size, mel_bins, fmin, fmax, classes_num): super(PVT_2layer, self).__init__() window = 'hann' center = True pad_mode = 'reflect' ref = 1.0 amin = 1e-10 top_db = None # Spectrogram extractor self.spectrogram_extractor = Spectrogram(n_fft=window_size, hop_length=hop_size, win_length=window_size, window=window, center=center, pad_mode=pad_mode, freeze_parameters=True) # Logmel feature extractor self.logmel_extractor = LogmelFilterBank(sr=sample_rate, n_fft=window_size, n_mels=mel_bins, fmin=fmin, fmax=fmax, ref=ref, amin=amin, top_db=top_db, freeze_parameters=True) self.time_shift = TimeShift(0, 10) # Spec augmenter self.spec_augmenter = SpecAugmentation(time_drop_width=64, time_stripes_num=2, freq_drop_width=8, freq_stripes_num=2) self.bn0 = nn.BatchNorm2d(64) self.pvt_transformer = PyramidVisionTransformerV2(tdim=1001, fdim=64, patch_size=7, stride=4, in_chans=1, num_classes=classes_num, embed_dims=[64, 128], depths=[3, 4], num_heads=[1, 2], mlp_ratios=[8, 8], qkv_bias=True, qk_scale=None, drop_rate=0.0, drop_path_rate=0.1, sr_ratios=[8, 4], norm_layer=partial(nn.LayerNorm, eps=1e-6), num_stages=2, pretrained='https://github.com/whai362/PVT/releases/download/v2/pvt_v2_b2.pth' ) #self.temp_pool = LinearSoftPool() self.avgpool = nn.AdaptiveAvgPool1d(1) self.fc_audioset = nn.Linear(128, classes_num, bias=True) self.init_weights() def init_weights(self): init_bn(self.bn0) init_layer(self.fc_audioset) def forward(self, input, mixup_lambda=None): """Input: (batch_size, times_steps, freq_bins)""" interpolate_ratio = 8 x = self.spectrogram_extractor(input) # (batch_size, 1, time_steps, freq_bins) x = self.logmel_extractor(x) # (batch_size, 1, time_steps, mel_bins) frames_num = x.shape[2] x = x.transpose(1, 3) x = self.bn0(x) x = x.transpose(1, 3) if self.training: x = self.time_shift(x) x = self.spec_augmenter(x) # Mixup on spectrogram if self.training and mixup_lambda is not None: x = do_mixup(x, mixup_lambda) #print(x.shape) #torch.Size([10, 1, 1001, 64]) x = self.pvt_transformer(x) #print(x.shape) #torch.Size([10, 800, 128]) x = torch.mean(x, dim=3) x = x.transpose(1, 2).contiguous() framewise_output = torch.sigmoid(self.fc_audioset(x)) #clipwise_output = torch.mean(framewise_output, dim=1) #clipwise_output = self.temp_pool(x, framewise_output).clamp(1e-7, 1.).squeeze(1) x = framewise_output.transpose(1, 2).contiguous() x = self.avgpool(x) clipwise_output = torch.flatten(x, 1) #print(framewise_output.shape) #torch.Size([10, 100, 17]) framewise_output = interpolate(framewise_output, interpolate_ratio) #framewise_output = framewise_output[:,:1000,:] #framewise_output = pad_framewise_output(framewise_output, frames_num) output_dict = {'framewise_output': framewise_output, 'clipwise_output': clipwise_output} return output_dict class PVT_lr(nn.Module): def __init__(self, sample_rate, window_size, hop_size, mel_bins, fmin, fmax, classes_num): super(PVT_lr, self).__init__() window = 'hann' center = True pad_mode = 'reflect' ref = 1.0 amin = 1e-10 top_db = None # Spectrogram extractor self.spectrogram_extractor = Spectrogram(n_fft=window_size, hop_length=hop_size, win_length=window_size, window=window, center=center, pad_mode=pad_mode, freeze_parameters=True) # Logmel feature extractor self.logmel_extractor = LogmelFilterBank(sr=sample_rate, n_fft=window_size, n_mels=mel_bins, fmin=fmin, fmax=fmax, ref=ref, amin=amin, top_db=top_db, freeze_parameters=True) self.time_shift = TimeShift(0, 10) # Spec augmenter self.spec_augmenter = SpecAugmentation(time_drop_width=64, time_stripes_num=2, freq_drop_width=8, freq_stripes_num=2) self.bn0 = nn.BatchNorm2d(64) self.pvt_transformer = PyramidVisionTransformerV2(tdim=1001, fdim=64, patch_size=7, stride=4, in_chans=1, num_classes=classes_num, embed_dims=[64, 128, 320, 512], depths=[3, 4, 6, 3], num_heads=[1, 2, 5, 8], mlp_ratios=[8, 8, 4, 4], qkv_bias=True, qk_scale=None, drop_rate=0.0, drop_path_rate=0.1, sr_ratios=[8, 4, 2, 1], norm_layer=partial(nn.LayerNorm, eps=1e-6), num_stages=4, pretrained='https://github.com/whai362/PVT/releases/download/v2/pvt_v2_b2.pth' ) self.temp_pool = LinearSoftPool() self.fc_audioset = nn.Linear(512, classes_num, bias=True) self.init_weights() def init_weights(self): init_bn(self.bn0) init_layer(self.fc_audioset) def forward(self, input, mixup_lambda=None): """Input: (batch_size, times_steps, freq_bins)""" interpolate_ratio = 32 x = self.spectrogram_extractor(input) # (batch_size, 1, time_steps, freq_bins) x = self.logmel_extractor(x) # (batch_size, 1, time_steps, mel_bins) frames_num = x.shape[2] x = x.transpose(1, 3) x = self.bn0(x) x = x.transpose(1, 3) if self.training: x = self.time_shift(x) x = self.spec_augmenter(x) # Mixup on spectrogram if self.training and mixup_lambda is not None: x = do_mixup(x, mixup_lambda) #print(x.shape) #torch.Size([10, 1, 1001, 64]) x = self.pvt_transformer(x) #print(x.shape) #torch.Size([10, 800, 128]) x = torch.mean(x, dim=3) x = x.transpose(1, 2).contiguous() framewise_output = torch.sigmoid(self.fc_audioset(x)) clipwise_output = self.temp_pool(x, framewise_output).clamp(1e-7, 1.).squeeze(1) #print(framewise_output.shape) #torch.Size([10, 100, 17]) framewise_output = interpolate(framewise_output, interpolate_ratio) #framewise_output = framewise_output[:,:1000,:] #framewise_output = pad_framewise_output(framewise_output, frames_num) output_dict = {'framewise_output': framewise_output, 'clipwise_output': clipwise_output} return output_dict class PVT_nopretrain(nn.Module): def __init__(self, sample_rate, window_size, hop_size, mel_bins, fmin, fmax, classes_num): super(PVT_nopretrain, self).__init__() window = 'hann' center = True pad_mode = 'reflect' ref = 1.0 amin = 1e-10 top_db = None # Spectrogram extractor self.spectrogram_extractor = Spectrogram(n_fft=window_size, hop_length=hop_size, win_length=window_size, window=window, center=center, pad_mode=pad_mode, freeze_parameters=True) # Logmel feature extractor self.logmel_extractor = LogmelFilterBank(sr=sample_rate, n_fft=window_size, n_mels=mel_bins, fmin=fmin, fmax=fmax, ref=ref, amin=amin, top_db=top_db, freeze_parameters=True) self.time_shift = TimeShift(0, 10) # Spec augmenter self.spec_augmenter = SpecAugmentation(time_drop_width=64, time_stripes_num=2, freq_drop_width=8, freq_stripes_num=2) self.bn0 = nn.BatchNorm2d(64) self.pvt_transformer = PyramidVisionTransformerV2(tdim=1001, fdim=64, patch_size=7, stride=4, in_chans=1, num_classes=classes_num, embed_dims=[64, 128, 320, 512], depths=[3, 4, 6, 3], num_heads=[1, 2, 5, 8], mlp_ratios=[8, 8, 4, 4], qkv_bias=True, qk_scale=None, drop_rate=0.0, drop_path_rate=0.1, sr_ratios=[8, 4, 2, 1], norm_layer=partial(nn.LayerNorm, eps=1e-6), num_stages=4, #pretrained='https://github.com/whai362/PVT/releases/download/v2/pvt_v2_b2.pth' ) self.temp_pool = LinearSoftPool() self.fc_audioset = nn.Linear(512, classes_num, bias=True) self.init_weights() def init_weights(self): init_bn(self.bn0) init_layer(self.fc_audioset) def forward(self, input, mixup_lambda=None): """Input: (batch_size, times_steps, freq_bins)""" interpolate_ratio = 32 x = self.spectrogram_extractor(input) # (batch_size, 1, time_steps, freq_bins) x = self.logmel_extractor(x) # (batch_size, 1, time_steps, mel_bins) frames_num = x.shape[2] x = x.transpose(1, 3) x = self.bn0(x) x = x.transpose(1, 3) if self.training: x = self.time_shift(x) x = self.spec_augmenter(x) # Mixup on spectrogram if self.training and mixup_lambda is not None: x = do_mixup(x, mixup_lambda) #print(x.shape) #torch.Size([10, 1, 1001, 64]) x = self.pvt_transformer(x) #print(x.shape) #torch.Size([10, 800, 128]) x = torch.mean(x, dim=3) x = x.transpose(1, 2).contiguous() framewise_output = torch.sigmoid(self.fc_audioset(x)) clipwise_output = self.temp_pool(x, framewise_output).clamp(1e-7, 1.).squeeze(1) #print(framewise_output.shape) #torch.Size([10, 100, 17]) framewise_output = interpolate(framewise_output, interpolate_ratio) framewise_output = framewise_output[:,:1000,:] #framewise_output = pad_framewise_output(framewise_output, frames_num) output_dict = {'framewise_output': framewise_output, 'clipwise_output': clipwise_output} return output_dict class Mlp(nn.Module): def __init__(self, in_features, hidden_features=None, out_features=None, act_layer=nn.GELU, drop=0., linear=False): super().__init__() out_features = out_features or in_features hidden_features = hidden_features or in_features self.fc1 = nn.Linear(in_features, hidden_features) self.dwconv = DWConv(hidden_features) self.act = act_layer() self.fc2 = nn.Linear(hidden_features, out_features) self.drop = nn.Dropout(drop) self.linear = linear if self.linear: self.relu = nn.ReLU() self.apply(self._init_weights) def _init_weights(self, m): if isinstance(m, nn.Linear): trunc_normal_(m.weight, std=.02) if isinstance(m, nn.Linear) and m.bias is not None: nn.init.constant_(m.bias, 0) elif isinstance(m, nn.LayerNorm): nn.init.constant_(m.bias, 0) nn.init.constant_(m.weight, 1.0) elif isinstance(m, nn.Conv2d): fan_out = m.kernel_size[0] * m.kernel_size[1] * m.out_channels fan_out //= m.groups m.weight.data.normal_(0, math.sqrt(2.0 / fan_out)) if m.bias is not None: m.bias.data.zero_() def forward(self, x, H, W): x = self.fc1(x) if self.linear: x = self.relu(x) x = self.dwconv(x, H, W) x = self.act(x) x = self.drop(x) x = self.fc2(x) x = self.drop(x) return x class Attention(nn.Module): def __init__(self, dim, num_heads=8, qkv_bias=False, qk_scale=None, attn_drop=0., proj_drop=0., sr_ratio=1, linear=False): super().__init__() assert dim % num_heads == 0, f"dim {dim} should be divided by num_heads {num_heads}." self.dim = dim self.num_heads = num_heads head_dim = dim // num_heads self.scale = qk_scale or head_dim ** -0.5 self.q = nn.Linear(dim, dim, bias=qkv_bias) self.kv = nn.Linear(dim, dim * 2, bias=qkv_bias) self.attn_drop = nn.Dropout(attn_drop) self.proj = nn.Linear(dim, dim) self.proj_drop = nn.Dropout(proj_drop) self.linear = linear self.sr_ratio = sr_ratio if not linear: if sr_ratio > 1: self.sr = nn.Conv2d(dim, dim, kernel_size=sr_ratio, stride=sr_ratio) self.norm = nn.LayerNorm(dim) else: self.pool = nn.AdaptiveAvgPool2d(7) self.sr = nn.Conv2d(dim, dim, kernel_size=1, stride=1) self.norm = nn.LayerNorm(dim) self.act = nn.GELU() self.apply(self._init_weights) def _init_weights(self, m): if isinstance(m, nn.Linear): trunc_normal_(m.weight, std=.02) if isinstance(m, nn.Linear) and m.bias is not None: nn.init.constant_(m.bias, 0) elif isinstance(m, nn.LayerNorm): nn.init.constant_(m.bias, 0) nn.init.constant_(m.weight, 1.0) elif isinstance(m, nn.Conv2d): fan_out = m.kernel_size[0] * m.kernel_size[1] * m.out_channels fan_out //= m.groups m.weight.data.normal_(0, math.sqrt(2.0 / fan_out)) if m.bias is not None: m.bias.data.zero_() def forward(self, x, H, W): B, N, C = x.shape q = self.q(x).reshape(B, N, self.num_heads, C // self.num_heads).permute(0, 2, 1, 3) if not self.linear: if self.sr_ratio > 1: x_ = x.permute(0, 2, 1).reshape(B, C, H, W) x_ = self.sr(x_).reshape(B, C, -1).permute(0, 2, 1) x_ = self.norm(x_) kv = self.kv(x_).reshape(B, -1, 2, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4) else: kv = self.kv(x).reshape(B, -1, 2, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4) else: x_ = x.permute(0, 2, 1).reshape(B, C, H, W) x_ = self.sr(self.pool(x_)).reshape(B, C, -1).permute(0, 2, 1) x_ = self.norm(x_) x_ = self.act(x_) kv = self.kv(x_).reshape(B, -1, 2, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4) k, v = kv[0], kv[1] attn = (q @ k.transpose(-2, -1)) * self.scale attn = attn.softmax(dim=-1) attn = self.attn_drop(attn) x = (attn @ v).transpose(1, 2).reshape(B, N, C) x = self.proj(x) x = self.proj_drop(x) return x class Pooling(nn.Module): """ Implementation of pooling for PoolFormer --pool_size: pooling size """ def __init__(self, pool_size=3): super().__init__() self.pool = nn.AvgPool2d( pool_size, stride=1, padding=pool_size//2, count_include_pad=False) def forward(self, x): return self.pool(x) - x class Block(nn.Module): def __init__(self, dim, num_heads, mlp_ratio=4., qkv_bias=False, qk_scale=None, drop=0., attn_drop=0., drop_path=0., act_layer=nn.GELU, norm_layer=nn.LayerNorm, sr_ratio=1, linear=False): super().__init__() self.norm1 = norm_layer(dim) self.attn = Attention( dim, num_heads=num_heads, qkv_bias=qkv_bias, qk_scale=qk_scale, attn_drop=attn_drop, proj_drop=drop, sr_ratio=sr_ratio, linear=linear) #self.norm3 = norm_layer(dim) #self.token_mixer = Pooling(pool_size=3) # NOTE: drop path for stochastic depth, we shall see if this is better than dropout here self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity() self.norm2 = norm_layer(dim) mlp_hidden_dim = int(dim * mlp_ratio) self.mlp = Mlp(in_features=dim, hidden_features=mlp_hidden_dim, act_layer=act_layer, drop=drop, linear=linear) self.apply(self._init_weights) def _init_weights(self, m): if isinstance(m, nn.Linear): trunc_normal_(m.weight, std=.02) if isinstance(m, nn.Linear) and m.bias is not None: nn.init.constant_(m.bias, 0) elif isinstance(m, nn.LayerNorm): nn.init.constant_(m.bias, 0) nn.init.constant_(m.weight, 1.0) elif isinstance(m, nn.Conv2d): fan_out = m.kernel_size[0] * m.kernel_size[1] * m.out_channels fan_out //= m.groups m.weight.data.normal_(0, math.sqrt(2.0 / fan_out)) if m.bias is not None: m.bias.data.zero_() def forward(self, x, H, W): x = x + self.drop_path(self.attn(self.norm1(x), H, W)) x = x + self.drop_path(self.mlp(self.norm2(x), H, W)) return x class OverlapPatchEmbed(nn.Module): """ Image to Patch Embedding """ def __init__(self, tdim, fdim, patch_size=7, stride=4, in_chans=3, embed_dim=768): super().__init__() img_size = (tdim, fdim) patch_size = to_2tuple(patch_size) self.img_size = img_size self.patch_size = patch_size self.H, self.W = img_size[0] // stride, img_size[1] // stride self.num_patches = self.H * self.W self.proj = nn.Conv2d(in_chans, embed_dim, kernel_size=patch_size, stride=stride, padding=(patch_size[0] // 3, patch_size[1] // 3)) self.norm = nn.LayerNorm(embed_dim) self.apply(self._init_weights) def _init_weights(self, m): if isinstance(m, nn.Linear): trunc_normal_(m.weight, std=.02) if isinstance(m, nn.Linear) and m.bias is not None: nn.init.constant_(m.bias, 0) elif isinstance(m, nn.LayerNorm): nn.init.constant_(m.bias, 0) nn.init.constant_(m.weight, 1.0) elif isinstance(m, nn.Conv2d): fan_out = m.kernel_size[0] * m.kernel_size[1] * m.out_channels fan_out //= m.groups m.weight.data.normal_(0, math.sqrt(2.0 / fan_out)) if m.bias is not None: m.bias.data.zero_() def forward(self, x): x = self.proj(x) _, _, H, W = x.shape x = x.flatten(2).transpose(1, 2) x = self.norm(x) return x, H, W class PyramidVisionTransformerV2(nn.Module): def __init__(self, tdim=1001, fdim=64, patch_size=16, stride=4, in_chans=3, num_classes=1000, embed_dims=[64, 128, 256, 512], num_heads=[1, 2, 4, 8], mlp_ratios=[4, 4, 4, 4], qkv_bias=False, qk_scale=None, drop_rate=0., attn_drop_rate=0., drop_path_rate=0.1, norm_layer=partial(nn.LayerNorm, eps=1e-6), depths=[3, 4, 6, 3], sr_ratios=[8, 4, 2, 1], num_stages=2, linear=False, pretrained=None): super().__init__() # self.num_classes = num_classes self.depths = depths self.num_stages = num_stages self.linear = linear dpr = [x.item() for x in torch.linspace(0, drop_path_rate, sum(depths))] # stochastic depth decay rule cur = 0 for i in range(num_stages): patch_embed = OverlapPatchEmbed(tdim=tdim if i == 0 else tdim // (2 ** (i + 1)), fdim=fdim if i == 0 else tdim // (2 ** (i + 1)), patch_size=7 if i == 0 else 3, stride=stride if i == 0 else 2, in_chans=in_chans if i == 0 else embed_dims[i - 1], embed_dim=embed_dims[i]) block = nn.ModuleList([Block( dim=embed_dims[i], num_heads=num_heads[i], mlp_ratio=mlp_ratios[i], qkv_bias=qkv_bias, qk_scale=qk_scale, drop=drop_rate, attn_drop=attn_drop_rate, drop_path=dpr[cur + j], norm_layer=norm_layer, sr_ratio=sr_ratios[i], linear=linear) for j in range(depths[i])]) norm = norm_layer(embed_dims[i]) cur += depths[i] setattr(self, f"patch_embed{i + 1}", patch_embed) setattr(self, f"block{i + 1}", block) setattr(self, f"norm{i + 1}", norm) #self.n = nn.Linear(125, 250, bias=True) # classification head # self.head = nn.Linear(embed_dims[3], num_classes) if num_classes > 0 else nn.Identity() self.apply(self._init_weights) self.init_weights(pretrained) def _init_weights(self, m): if isinstance(m, nn.Linear): trunc_normal_(m.weight, std=.02) if isinstance(m, nn.Linear) and m.bias is not None: nn.init.constant_(m.bias, 0) elif isinstance(m, nn.LayerNorm): nn.init.constant_(m.bias, 0) nn.init.constant_(m.weight, 1.0) elif isinstance(m, nn.Conv2d): fan_out = m.kernel_size[0] * m.kernel_size[1] * m.out_channels fan_out //= m.groups m.weight.data.normal_(0, math.sqrt(2.0 / fan_out)) if m.bias is not None: m.bias.data.zero_() def init_weights(self, pretrained=None): if isinstance(pretrained, str): logger = get_root_logger() load_checkpoint(self, pretrained, map_location='cpu', strict=False, logger=logger) def freeze_patch_emb(self): self.patch_embed1.requires_grad = False @torch.jit.ignore def no_weight_decay(self): return {'pos_embed1', 'pos_embed2', 'pos_embed3', 'pos_embed4', 'cls_token'} # has pos_embed may be better def get_classifier(self): return self.head def reset_classifier(self, num_classes, global_pool=''): self.num_classes = num_classes self.head = nn.Linear(self.embed_dim, num_classes) if num_classes > 0 else nn.Identity() def forward_features(self, x): B = x.shape[0] for i in range(self.num_stages): patch_embed = getattr(self, f"patch_embed{i + 1}") block = getattr(self, f"block{i + 1}") norm = getattr(self, f"norm{i + 1}") x, H, W = patch_embed(x) #print(x.shape) for blk in block: x = blk(x, H, W) #print(x.shape) x = norm(x) #if i != self.num_stages - 1: x = x.reshape(B, H, W, -1).permute(0, 3, 1, 2).contiguous() #print(x.shape) return x def forward(self, x): x = self.forward_features(x) # x = self.head(x) return x class DWConv(nn.Module): def __init__(self, dim=768): super(DWConv, self).__init__() self.dwconv = nn.Conv2d(dim, dim, 3, 1, 1, bias=True, groups=dim) def forward(self, x, H, W): B, N, C = x.shape x = x.transpose(1, 2).view(B, C, H, W) x = self.dwconv(x) x = x.flatten(2).transpose(1, 2) return x def _conv_filter(state_dict, patch_size=16): """ convert patch embedding weight from manual patchify + linear proj to conv""" out_dict = {} for k, v in state_dict.items(): if 'patch_embed.proj.weight' in k: v = v.reshape((v.shape[0], 3, patch_size, patch_size)) out_dict[k] = v return out_dict