# Copyright (c) 2014-2021 Megvii Inc And Alibaba PAI-Teams. All rights reserved. import logging import math from abc import abstractmethod from distutils.version import LooseVersion import torch import torch.nn as nn import torch.nn.functional as F from easycv.framework.errors import KeyError, RuntimeError from easycv.models.backbones.network_blocks import BaseConv, DWConv from easycv.models.backbones.repvgg_yolox_backbone import RepVGGBlock from easycv.models.detection.utils import bboxes_iou from easycv.models.loss import YOLOX_IOULoss class YOLOXHead_Template(nn.Module): param_map = { 'nano': [0.33, 0.25], 'tiny': [0.33, 0.375], 's': [0.33, 0.5], 'm': [0.67, 0.75], 'l': [1.0, 1.0], 'x': [1.33, 1.25] } def __init__(self, num_classes=80, model_type='s', strides=[8, 16, 32], in_channels=[256, 512, 1024], act='silu', conv_type='conv', stage='CLOUD', obj_loss_type='BCE', reg_loss_type='giou', decode_in_inference=True, width=None): """ Args: num_classes (int): detection class numbers. width (float): model width. Default value: 1.0. strides (list): expanded strides. Default value: [8, 16, 32]. in_channels (list): model conv channels set. Default value: [256, 512, 1024]. act (str): activation type of conv. Defalut value: "silu". depthwise (bool): whether apply depthwise conv in conv branch. Default value: False. stage (str): model stage, distinguish edge head to cloud head. Default value: CLOUD. obj_loss_type (str): the loss function of the obj conf. Default value: BCE. reg_loss_type (str): the loss function of the box prediction. Default value: giou. """ super().__init__() if width is None and model_type in self.param_map: width = self.param_map[model_type][1] else: assert (width != None), 'Unknow model type must have a given width!' self.width = width self.n_anchors = 1 self.num_classes = num_classes self.stage = stage self.decode_in_inference = decode_in_inference # for deploy, set to False self.cls_convs = nn.ModuleList() self.reg_convs = nn.ModuleList() self.cls_preds = nn.ModuleList() self.reg_preds = nn.ModuleList() self.obj_preds = nn.ModuleList() self.stems = nn.ModuleList() default_conv_type_list = ['conv', 'dwconv', 'repconv'] if conv_type not in default_conv_type_list: logging.warning( 'YOLOX-PAI tood head conv_type must in [conv, dwconv, repconv], otherwise we use repconv as default' ) conv_type = 'repconv' if conv_type == 'conv': Conv = BaseConv if conv_type == 'dwconv': Conv = DWConv if conv_type == 'repconv': Conv = RepVGGBlock for i in range(len(in_channels)): self.stems.append( BaseConv( in_channels=int(in_channels[i] * width), out_channels=int(256 * width), ksize=1, stride=1, act=act, )) self.cls_convs.append( nn.Sequential(*[ Conv( in_channels=int(256 * width), out_channels=int(256 * width), ksize=3, stride=1, act=act, ), Conv( in_channels=int(256 * width), out_channels=int(256 * width), ksize=3, stride=1, act=act, ), ])) self.reg_convs.append( nn.Sequential(*[ Conv( in_channels=int(256 * width), out_channels=int(256 * width), ksize=3, stride=1, act=act, ), Conv( in_channels=int(256 * width), out_channels=int(256 * width), ksize=3, stride=1, act=act, ), ])) self.cls_preds.append( nn.Conv2d( in_channels=int(256 * width), out_channels=self.n_anchors * self.num_classes, kernel_size=1, stride=1, padding=0, )) self.reg_preds.append( nn.Conv2d( in_channels=int(256 * width), out_channels=4, kernel_size=1, stride=1, padding=0, )) self.obj_preds.append( nn.Conv2d( in_channels=int(256 * width), out_channels=self.n_anchors * 1, kernel_size=1, stride=1, padding=0, )) self.bcewithlog_loss = nn.BCEWithLogitsLoss(reduction='none') self.use_l1 = False self.l1_loss = nn.L1Loss(reduction='none') self.iou_loss = YOLOX_IOULoss( reduction='none', loss_type=reg_loss_type) self.obj_loss_type = obj_loss_type if obj_loss_type == 'BCE': self.obj_loss = nn.BCEWithLogitsLoss(reduction='none') else: raise KeyError('Undefined loss type: {}'.format(obj_loss_type)) self.strides = strides self.grids = [torch.zeros(1)] * len(in_channels) def initialize_biases(self, prior_prob): for conv in self.cls_preds: b = conv.bias.view(self.n_anchors, -1) b.data.fill_(-math.log((1 - prior_prob) / prior_prob)) conv.bias = torch.nn.Parameter(b.view(-1), requires_grad=True) for conv in self.obj_preds: b = conv.bias.view(self.n_anchors, -1) b.data.fill_(-math.log((1 - prior_prob) / prior_prob)) conv.bias = torch.nn.Parameter(b.view(-1), requires_grad=True) def get_nmsboxes_num(self, img_scale=(640, 640)): """ Count all Yolox NMS box with img_scale and head stride config """ assert ( len(img_scale) == 2 ), 'Export YoloX predictor config contains img_scale must be (int, int) tuple!' total_box_count = 0 for stride in self.strides: total_box_count += (img_scale[0] / stride) * ( img_scale[1] / stride) return total_box_count @abstractmethod def forward(self, xin, labels=None, imgs=None): pass def get_output_and_grid(self, output, k, stride, dtype): grid = self.grids[k] batch_size = output.shape[0] n_ch = 5 + self.num_classes hsize, wsize = output.shape[-2:] if grid.shape[2:4] != output.shape[2:4]: yv, xv = torch.meshgrid([torch.arange(hsize), torch.arange(wsize)]) grid = torch.stack((xv, yv), 2).view(1, 1, hsize, wsize, 2).type(dtype) self.grids[k] = grid output = output.view(batch_size, self.n_anchors, n_ch, hsize, wsize) output = output.permute(0, 1, 3, 4, 2).reshape(batch_size, self.n_anchors * hsize * wsize, -1) grid = grid.view(1, -1, 2) output[..., :2] = (output[..., :2] + grid) * stride output[..., 2:4] = torch.exp(output[..., 2:4]) * stride return output, grid def decode_outputs(self, outputs, dtype): grids = [] strides = [] for (hsize, wsize), stride in zip(self.hw, self.strides): yv, xv = torch.meshgrid([torch.arange(hsize), torch.arange(wsize)]) grid = torch.stack((xv, yv), 2).view(1, -1, 2) grids.append(grid) shape = grid.shape[:2] strides.append(torch.full((*shape, 1), stride, dtype=torch.int)) grids = torch.cat(grids, dim=1).type(dtype) strides = torch.cat(strides, dim=1).type(dtype) outputs[..., :2] = (outputs[..., :2] + grids) * strides outputs[..., 2:4] = torch.exp(outputs[..., 2:4]) * strides return outputs def get_losses( self, imgs, x_shifts, y_shifts, expanded_strides, labels, outputs, origin_preds, dtype, ): bbox_preds = outputs[:, :, :4] # [batch, n_anchors_all, 4] obj_preds = outputs[:, :, 4].unsqueeze(-1) # [batch, n_anchors_all, 1] cls_preds = outputs[:, :, 5:] # [batch, n_anchors_all, n_cls] # calculate targets nlabel = (labels.sum(dim=2) > 0).sum(dim=1) # number of objects total_num_anchors = outputs.shape[1] x_shifts = torch.cat(x_shifts, 1) # [1, n_anchors_all] y_shifts = torch.cat(y_shifts, 1) # [1, n_anchors_all] expanded_strides = torch.cat(expanded_strides, 1) if self.use_l1: origin_preds = torch.cat(origin_preds, 1) cls_targets = [] reg_targets = [] l1_targets = [] obj_targets = [] fg_masks = [] num_fg = 0.0 num_gts = 0.0 for batch_idx in range(outputs.shape[0]): num_gt = int(nlabel[batch_idx]) num_gts += num_gt if num_gt == 0: cls_target = outputs.new_zeros((0, self.num_classes)) reg_target = outputs.new_zeros((0, 4)) l1_target = outputs.new_zeros((0, 4)) obj_target = outputs.new_zeros((total_num_anchors, 1)) fg_mask = outputs.new_zeros(total_num_anchors).bool() else: gt_bboxes_per_image = labels[batch_idx, :num_gt, 1:5] gt_classes = labels[batch_idx, :num_gt, 0] bboxes_preds_per_image = bbox_preds[batch_idx] try: ( gt_matched_classes, fg_mask, pred_ious_this_matching, matched_gt_inds, num_fg_img, ) = self.get_assignments( # noqa batch_idx, num_gt, total_num_anchors, gt_bboxes_per_image, gt_classes, bboxes_preds_per_image, expanded_strides, x_shifts, y_shifts, cls_preds, bbox_preds, obj_preds, labels, imgs, ) except RuntimeError: logging.error( 'OOM RuntimeError is raised due to the huge memory cost during label assignment. \ CPU mode is applied in this batch. If you want to avoid this issue, \ try to reduce the batch size or image size.') torch.cuda.empty_cache() ( gt_matched_classes, fg_mask, pred_ious_this_matching, matched_gt_inds, num_fg_img, ) = self.get_assignments( # noqa batch_idx, num_gt, total_num_anchors, gt_bboxes_per_image, gt_classes, bboxes_preds_per_image, expanded_strides, x_shifts, y_shifts, cls_preds, bbox_preds, obj_preds, labels, imgs, 'cpu', ) torch.cuda.empty_cache() num_fg += num_fg_img cls_target = F.one_hot( gt_matched_classes.to(torch.int64), self.num_classes) * pred_ious_this_matching.unsqueeze(-1) obj_target = fg_mask.unsqueeze(-1) reg_target = gt_bboxes_per_image[matched_gt_inds] if self.use_l1: l1_target = self.get_l1_target( outputs.new_zeros((num_fg_img, 4)), gt_bboxes_per_image[matched_gt_inds], expanded_strides[0][fg_mask], x_shifts=x_shifts[0][fg_mask], y_shifts=y_shifts[0][fg_mask], ) cls_targets.append(cls_target) reg_targets.append(reg_target) obj_targets.append(obj_target.to(dtype)) fg_masks.append(fg_mask) if self.use_l1: l1_targets.append(l1_target) cls_targets = torch.cat(cls_targets, 0) reg_targets = torch.cat(reg_targets, 0) obj_targets = torch.cat(obj_targets, 0) fg_masks = torch.cat(fg_masks, 0) if self.use_l1: l1_targets = torch.cat(l1_targets, 0) num_fg = max(num_fg, 1) loss_iou = (self.iou_loss( bbox_preds.view(-1, 4)[fg_masks], reg_targets)).sum() / num_fg # loss_iou1 = (self.iou_loss1( # bbox_preds.view(-1, 4)[fg_masks], reg_targets,xyxy=False)).sum() / num_fg loss_obj = (self.obj_loss(obj_preds.view(-1, 1), obj_targets)).sum() / num_fg loss_cls = (self.bcewithlog_loss( cls_preds.view(-1, self.num_classes)[fg_masks], cls_targets)).sum() / num_fg if self.use_l1: loss_l1 = (self.l1_loss( origin_preds.view(-1, 4)[fg_masks], l1_targets)).sum() / num_fg else: loss_l1 = 0.0 reg_weight = 5.0 loss = reg_weight * loss_iou + loss_obj + loss_cls + loss_l1 return ( loss, reg_weight * loss_iou, loss_obj, loss_cls, loss_l1, num_fg / max(num_gts, 1), ) def get_l1_target(self, l1_target, gt, stride, x_shifts, y_shifts, eps=1e-8): l1_target[:, 0] = gt[:, 0] / stride - x_shifts l1_target[:, 1] = gt[:, 1] / stride - y_shifts l1_target[:, 2] = torch.log(gt[:, 2] / stride + eps) l1_target[:, 3] = torch.log(gt[:, 3] / stride + eps) return l1_target @torch.no_grad() def get_assignments( self, batch_idx, num_gt, total_num_anchors, gt_bboxes_per_image, gt_classes, bboxes_preds_per_image, expanded_strides, x_shifts, y_shifts, cls_preds, bbox_preds, obj_preds, labels, imgs, mode='gpu', ): if mode == 'cpu': print('------------CPU Mode for This Batch-------------') gt_bboxes_per_image = gt_bboxes_per_image.cpu().float() bboxes_preds_per_image = bboxes_preds_per_image.cpu().float() gt_classes = gt_classes.cpu().float() expanded_strides = expanded_strides.cpu().float() x_shifts = x_shifts.cpu() y_shifts = y_shifts.cpu() fg_mask, is_in_boxes_and_center = self.get_in_boxes_info( gt_bboxes_per_image, expanded_strides, x_shifts, y_shifts, total_num_anchors, num_gt, ) # reference to: https://github.com/Megvii-BaseDetection/YOLOX/pull/811 # NOTE: Fix `selected index k out of range` num_pos_anchors: int = fg_mask.sum().item( ) # number of positive anchors if num_pos_anchors == 0: gt_matched_classes = torch.zeros(0, device=fg_mask.device).long() pred_ious_this_matching = torch.rand(0, device=fg_mask.device) matched_gt_inds = gt_matched_classes num_fg = num_pos_anchors if mode == 'cpu': gt_matched_classes = gt_matched_classes.cuda() fg_mask = fg_mask.cuda() pred_ious_this_matching = pred_ious_this_matching.cuda() matched_gt_inds = matched_gt_inds.cuda() num_fg = num_fg.cuda() return ( gt_matched_classes, fg_mask, pred_ious_this_matching, matched_gt_inds, num_fg, ) bboxes_preds_per_image = bboxes_preds_per_image[fg_mask] cls_preds_ = cls_preds[batch_idx][fg_mask] obj_preds_ = obj_preds[batch_idx][fg_mask] num_in_boxes_anchor = bboxes_preds_per_image.shape[0] if mode == 'cpu': gt_bboxes_per_image = gt_bboxes_per_image.cpu() bboxes_preds_per_image = bboxes_preds_per_image.cpu() pair_wise_ious = bboxes_iou(gt_bboxes_per_image, bboxes_preds_per_image, False) if (torch.isnan(pair_wise_ious.max())): pair_wise_ious = bboxes_iou(gt_bboxes_per_image, bboxes_preds_per_image, False) gt_cls_per_image = ( F.one_hot(gt_classes.to(torch.int64), self.num_classes).float().unsqueeze(1).repeat( 1, num_in_boxes_anchor, 1)) pair_wise_ious_loss = -torch.log(pair_wise_ious + 1e-8) if mode == 'cpu': cls_preds_, obj_preds_ = cls_preds_.cpu(), obj_preds_.cpu() if LooseVersion(torch.__version__) >= LooseVersion('1.6.0'): with torch.cuda.amp.autocast(enabled=False): cls_preds_ = ( cls_preds_.float().unsqueeze(0).repeat(num_gt, 1, 1).sigmoid_() * obj_preds_.float().unsqueeze(0).repeat(num_gt, 1, 1).sigmoid_()) pair_wise_cls_loss = F.binary_cross_entropy( cls_preds_.sqrt_(), gt_cls_per_image, reduction='none').sum(-1) else: cls_preds_ = ( cls_preds_.float().unsqueeze(0).repeat(num_gt, 1, 1).sigmoid_() * obj_preds_.float().unsqueeze(0).repeat(num_gt, 1, 1).sigmoid_()) pair_wise_cls_loss = F.binary_cross_entropy( cls_preds_.sqrt_(), gt_cls_per_image, reduction='none').sum(-1) del cls_preds_ cost = ( pair_wise_cls_loss + 3.0 * pair_wise_ious_loss + 100000.0 * (~is_in_boxes_and_center)) ( num_fg, gt_matched_classes, pred_ious_this_matching, matched_gt_inds, ) = self.dynamic_k_matching(cost, pair_wise_ious, gt_classes, num_gt, fg_mask) del pair_wise_cls_loss, cost, pair_wise_ious, pair_wise_ious_loss if mode == 'cpu': gt_matched_classes = gt_matched_classes.cuda() fg_mask = fg_mask.cuda() pred_ious_this_matching = pred_ious_this_matching.cuda() matched_gt_inds = matched_gt_inds.cuda() return ( gt_matched_classes, fg_mask, pred_ious_this_matching, matched_gt_inds, num_fg, ) def get_in_boxes_info( self, gt_bboxes_per_image, expanded_strides, x_shifts, y_shifts, total_num_anchors, num_gt, ): expanded_strides_per_image = expanded_strides[0] x_shifts_per_image = x_shifts[0] * expanded_strides_per_image y_shifts_per_image = y_shifts[0] * expanded_strides_per_image x_centers_per_image = ( (x_shifts_per_image + 0.5 * expanded_strides_per_image).unsqueeze(0).repeat(num_gt, 1) ) # [n_anchor] -> [n_gt, n_anchor] y_centers_per_image = ( (y_shifts_per_image + 0.5 * expanded_strides_per_image).unsqueeze(0).repeat(num_gt, 1)) gt_bboxes_per_image_l = ( (gt_bboxes_per_image[:, 0] - 0.5 * gt_bboxes_per_image[:, 2]).unsqueeze(1).repeat( 1, total_num_anchors)) gt_bboxes_per_image_r = ( (gt_bboxes_per_image[:, 0] + 0.5 * gt_bboxes_per_image[:, 2]).unsqueeze(1).repeat( 1, total_num_anchors)) gt_bboxes_per_image_t = ( (gt_bboxes_per_image[:, 1] - 0.5 * gt_bboxes_per_image[:, 3]).unsqueeze(1).repeat( 1, total_num_anchors)) gt_bboxes_per_image_b = ( (gt_bboxes_per_image[:, 1] + 0.5 * gt_bboxes_per_image[:, 3]).unsqueeze(1).repeat( 1, total_num_anchors)) b_l = x_centers_per_image - gt_bboxes_per_image_l b_r = gt_bboxes_per_image_r - x_centers_per_image b_t = y_centers_per_image - gt_bboxes_per_image_t b_b = gt_bboxes_per_image_b - y_centers_per_image bbox_deltas = torch.stack([b_l, b_t, b_r, b_b], 2) is_in_boxes = bbox_deltas.min(dim=-1).values > 0.0 is_in_boxes_all = is_in_boxes.sum(dim=0) > 0 # in fixed center center_radius = 2.5 gt_bboxes_per_image_l = ( gt_bboxes_per_image[:, 0]).unsqueeze(1).repeat( 1, total_num_anchors ) - center_radius * expanded_strides_per_image.unsqueeze(0) gt_bboxes_per_image_r = ( gt_bboxes_per_image[:, 0]).unsqueeze(1).repeat( 1, total_num_anchors ) + center_radius * expanded_strides_per_image.unsqueeze(0) gt_bboxes_per_image_t = ( gt_bboxes_per_image[:, 1]).unsqueeze(1).repeat( 1, total_num_anchors ) - center_radius * expanded_strides_per_image.unsqueeze(0) gt_bboxes_per_image_b = ( gt_bboxes_per_image[:, 1]).unsqueeze(1).repeat( 1, total_num_anchors ) + center_radius * expanded_strides_per_image.unsqueeze(0) c_l = x_centers_per_image - gt_bboxes_per_image_l c_r = gt_bboxes_per_image_r - x_centers_per_image c_t = y_centers_per_image - gt_bboxes_per_image_t c_b = gt_bboxes_per_image_b - y_centers_per_image center_deltas = torch.stack([c_l, c_t, c_r, c_b], 2) is_in_centers = center_deltas.min(dim=-1).values > 0.0 is_in_centers_all = is_in_centers.sum(dim=0) > 0 # in boxes and in centers is_in_boxes_anchor = is_in_boxes_all | is_in_centers_all is_in_boxes_and_center = ( is_in_boxes[:, is_in_boxes_anchor] & is_in_centers[:, is_in_boxes_anchor]) return is_in_boxes_anchor, is_in_boxes_and_center def dynamic_k_matching(self, cost, pair_wise_ious, gt_classes, num_gt, fg_mask): # Dynamic K # --------------------------------------------------------------- matching_matrix = torch.zeros_like(cost, dtype=torch.uint8) ious_in_boxes_matrix = pair_wise_ious n_candidate_k = min(10, ious_in_boxes_matrix.size(1)) topk_ious, _ = torch.topk(ious_in_boxes_matrix, n_candidate_k, dim=1) dynamic_ks = torch.clamp(topk_ious.sum(1).int(), min=1) dynamic_ks = dynamic_ks.tolist() for gt_idx in range(num_gt): _, pos_idx = torch.topk( cost[gt_idx], k=dynamic_ks[gt_idx], largest=False) matching_matrix[gt_idx][pos_idx] = 1 del topk_ious, dynamic_ks, pos_idx anchor_matching_gt = matching_matrix.sum(0) if (anchor_matching_gt > 1).sum() > 0: _, cost_argmin = torch.min(cost[:, anchor_matching_gt > 1], dim=0) matching_matrix[:, anchor_matching_gt > 1] *= 0 matching_matrix[cost_argmin, anchor_matching_gt > 1] = 1 fg_mask_inboxes = matching_matrix.sum(0) > 0 num_fg = fg_mask_inboxes.sum().item() fg_mask[fg_mask.clone()] = fg_mask_inboxes matched_gt_inds = matching_matrix[:, fg_mask_inboxes].argmax(0) gt_matched_classes = gt_classes[matched_gt_inds] pred_ious_this_matching = (matching_matrix * pair_wise_ious).sum(0)[fg_mask_inboxes] return num_fg, gt_matched_classes, pred_ious_this_matching, matched_gt_inds