@@ -1,3 +1,34 @@
+# BSD 3-Clause License
+#
+# Copyright (c) 2017 xxxx
+# All rights reserved.
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# * Redistributions of source code must retain the above copyright notice, this
+# list of conditions and the following disclaimer.
+#
+# * Redistributions in binary form must reproduce the above copyright notice,
+# this list of conditions and the following disclaimer in the documentation
+# and/or other materials provided with the distribution.
+#
+# * Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+# ============================================================================
import torch
import torch.nn as nn
import pickle
@@ -1,3 +1,34 @@
+# BSD 3-Clause License
+#
+# Copyright (c) 2017 xxxx
+# All rights reserved.
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# * Redistributions of source code must retain the above copyright notice, this
+# list of conditions and the following disclaimer.
+#
+# * Redistributions in binary form must reproduce the above copyright notice,
+# this list of conditions and the following disclaimer in the documentation
+# and/or other materials provided with the distribution.
+#
+# * Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+# ============================================================================
import os
import os.path as osp
import sys
@@ -53,17 +84,29 @@ class COCODetection(data.Dataset):
"""`MS Coco Detection <http://mscoco.org/dataset/#detections-challenge2016>`_ Dataset.
Args:
root (string): Root directory where images are downloaded to.
+ 这里即为image_path,数据集图像路径
+
set_name (string): Name of the specific set of COCO images.
+ 自定义，用于标记
+
transform (callable, optional): A function/transform that augments the
raw images`
+ 图像增强方法
+
target_transform (callable, optional): A function/transform that takes
in the target (bbox) and transforms it.
+ 将数据集中的目标检测框（bounding box）等封装为一个专门的数据结构
+
prep_crowds (bool): Whether or not to prepare crowds for the evaluation step.
+ ？
"""
def __init__(self, image_path, info_file, transform=None,
target_transform=None,
dataset_name='MS COCO', has_gt=True):
+ #has_gt中的gt代表ground truth，在监督学习中，数据是有标注的，以(x, t)的形式出现，其中x是输入数据，t是标注.
+ # 正确的t标注是ground truth， 错误的标记则不是。（也有人将所有标注数据都叫做ground truth）
+
# Do this here because we have too many things named COCO
from pycocotools.coco import COCO
@@ -71,14 +114,15 @@ class COCODetection(data.Dataset):
target_transform = COCOAnnotationTransform()
self.root = image_path
- self.coco = COCO(info_file)
-
+ self.coco = COCO(info_file) #加载标注文件, 其中COCO对象属性有1、anns ：所有标注；2、catToImgs：{种类:[图像list]}的映射字典；
+ # 3、cats：图片中包含物体的种类；4、dataset：内含信息、许可、标注等；5、imgToAnns：{图像:[标注list]}的映射字典；6、imgs：全部图像
+
self.ids = list(self.coco.imgToAnns.keys())
- if len(self.ids) == 0 or not has_gt:
+ if len(self.ids) == 0 or not has_gt: #显然这个if块内语句基本不可能执行
self.ids = list(self.coco.imgs.keys())
- self.transform = transform
- self.target_transform = COCOAnnotationTransform()
+ self.transform = transform #默认情况下，训练数据使用：SSDAugmentation，验证数据使用：baseTransform
+ self.target_transform = COCOAnnotationTransform() #82行的if块毫无意义
self.name = dataset_name
self.has_gt = has_gt
@@ -92,7 +136,7 @@ class COCODetection(data.Dataset):
target is the object returned by ``coco.loadAnns``.
"""
im, gt, masks, h, w, num_crowds = self.pull_item(index)
- return im, (gt, masks, num_crowds)
+ return (index, im), (gt, masks, num_crowds)
def __len__(self):
return len(self.ids)
@@ -1,3 +1,34 @@
+# BSD 3-Clause License
+#
+# Copyright (c) 2017 xxxx
+# All rights reserved.
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# * Redistributions of source code must retain the above copyright notice, this
+# list of conditions and the following disclaimer.
+#
+# * Redistributions in binary form must reproduce the above copyright notice,
+# this list of conditions and the following disclaimer in the documentation
+# and/or other materials provided with the distribution.
+#
+# * Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+# ============================================================================
from backbone import ResNetBackbone, VGGBackbone, ResNetBackboneGN, DarkNetBackbone
from math import sqrt
import torch
@@ -45,14 +76,14 @@ COCO_CLASSES = ('person', 'bicycle', 'car', 'motorcycle', 'airplane', 'bus',
COCO_LABEL_MAP = { 1: 1, 2: 2, 3: 3, 4: 4, 5: 5, 6: 6, 7: 7, 8: 8,
9: 9, 10: 10, 11: 11, 13: 12, 14: 13, 15: 14, 16: 15, 17: 16,
- 18: 17, 19: 18, 20: 19, 21: 20, 22: 21, 23: 22, 24: 23, 25: 24,
- 27: 25, 28: 26, 31: 27, 32: 28, 33: 29, 34: 30, 35: 31, 36: 32,
- 37: 33, 38: 34, 39: 35, 40: 36, 41: 37, 42: 38, 43: 39, 44: 40,
- 46: 41, 47: 42, 48: 43, 49: 44, 50: 45, 51: 46, 52: 47, 53: 48,
- 54: 49, 55: 50, 56: 51, 57: 52, 58: 53, 59: 54, 60: 55, 61: 56,
- 62: 57, 63: 58, 64: 59, 65: 60, 67: 61, 70: 62, 72: 63, 73: 64,
- 74: 65, 75: 66, 76: 67, 77: 68, 78: 69, 79: 70, 80: 71, 81: 72,
- 82: 73, 84: 74, 85: 75, 86: 76, 87: 77, 88: 78, 89: 79, 90: 80}
+ 18: 17, 19: 18, 20: 19, 21: 20, 22: 21, 23: 22, 24: 23, 25: 24,
+ 27: 25, 28: 26, 31: 27, 32: 28, 33: 29, 34: 30, 35: 31, 36: 32,
+ 37: 33, 38: 34, 39: 35, 40: 36, 41: 37, 42: 38, 43: 39, 44: 40,
+ 46: 41, 47: 42, 48: 43, 49: 44, 50: 45, 51: 46, 52: 47, 53: 48,
+ 54: 49, 55: 50, 56: 51, 57: 52, 58: 53, 59: 54, 60: 55, 61: 56,
+ 62: 57, 63: 58, 64: 59, 65: 60, 67: 61, 70: 62, 72: 63, 73: 64,
+ 74: 65, 75: 66, 76: 67, 77: 68, 78: 69, 79: 70, 80: 71, 81: 72,
+ 82: 73, 84: 74, 85: 75, 86: 76, 87: 77, 88: 78, 89: 79, 90: 80}
@@ -78,7 +109,7 @@ class Config(object):
"""
ret = Config(vars(self))
-
+
for key, val in new_config_dict.items():
ret.__setattr__(key, val)
@@ -94,7 +125,7 @@ class Config(object):
for key, val in new_config_dict.items():
self.__setattr__(key, val)
-
+
def print(self):
for k, v in vars(self).items():
print(k, ' = ', v)
@@ -109,11 +140,11 @@ dataset_base = Config({
'name': 'Base Dataset',
# Training images and annotations
- 'train_images': './data/coco/images/',
+ 'train_images': '/home/data/coco/images/',
'train_info': 'path_to_annotation_file',
# Validation images and annotations.
- 'valid_images': './data/coco/images/',
+ 'valid_images': '/home/data/coco/images/',
'valid_info': 'path_to_annotation_file',
# Whether or not to load GT. If this is False, eval.py quantitative evaluation won't work.
@@ -130,18 +161,18 @@ dataset_base = Config({
coco2014_dataset = dataset_base.copy({
'name': 'COCO 2014',
-
- 'train_info': './data/coco/annotations/instances_train2014.json',
- 'valid_info': './data/coco/annotations/instances_val2014.json',
+
+ 'train_info': '/home/data/coco/annotations/instances_train2014.json',
+ 'valid_info': '/home/data/coco/annotations/instances_val2014.json',
'label_map': COCO_LABEL_MAP
})
coco2017_dataset = dataset_base.copy({
'name': 'COCO 2017',
-
- 'train_info': './data/coco/annotations/instances_train2017.json',
- 'valid_info': './data/coco/annotations/instances_val2017.json',
+
+ 'train_info': '/home/data/coco/annotations/instances_train2017.json',
+ 'valid_info': '/home/data/coco/annotations/instances_val2017.json',
'label_map': COCO_LABEL_MAP
})
@@ -149,7 +180,7 @@ coco2017_dataset = dataset_base.copy({
coco2017_testdev_dataset = dataset_base.copy({
'name': 'COCO 2017 Test-Dev',
- 'valid_info': './data/coco/annotations/image_info_test-dev2017.json',
+ 'valid_info': '/home/data/coco/annotations/image_info_test-dev2017.json',
'has_gt': False,
'label_map': COCO_LABEL_MAP
@@ -165,7 +196,7 @@ pascal_sbd_dataset = dataset_base.copy({
'train_images': './data/sbd/img',
'valid_images': './data/sbd/img',
-
+
'train_info': './data/sbd/pascal_sbd_train.json',
'valid_info': './data/sbd/pascal_sbd_val.json',
@@ -520,7 +551,7 @@ coco_base_config = Config({
'use_focal_loss': False,
'focal_loss_alpha': 0.25,
'focal_loss_gamma': 2,
-
+
# The initial bias toward forground objects, as specified in the focal loss paper
'focal_loss_init_pi': 0.01,
@@ -582,7 +613,7 @@ coco_base_config = Config({
# Input image size.
'max_size': 300,
-
+
# Whether or not to do post processing on the cpu at test time
'force_cpu_nms': True,
@@ -612,7 +643,7 @@ coco_base_config = Config({
# Whether or not to use the predicted coordinate scheme from Yolo v2
'use_yolo_regressors': False,
-
+
# For training, bboxes are considered "positive" if their anchors have a 0.5 IoU overlap
# or greater with a ground truth box. If this is true, instead of using the anchor boxes
# for this IoU computation, the matching function will use the predicted bbox coordinates.
@@ -634,7 +665,7 @@ coco_base_config = Config({
# Do not crop out the mask with bbox but slide a convnet on the image-size mask,
# then use global pooling to get the final mask score
'use_maskiou': False,
-
+
# Archecture for the mask iou network. A (num_classes-1, 1, {}) layer is appended to the end.
'maskiou_net': [],
@@ -662,11 +693,11 @@ yolact_base_config = coco_base_config.copy({
# Image Size
'max_size': 550,
-
+
# Training params
'lr_steps': (280000, 600000, 700000, 750000),
'max_iter': 800000,
-
+
# Backbone Settings
'backbone': resnet101_backbone.copy({
'selected_layers': list(range(1, 4)),
@@ -727,7 +758,7 @@ yolact_darknet53_config = yolact_base_config.copy({
'backbone': darknet53_backbone.copy({
'selected_layers': list(range(2, 5)),
-
+
'pred_scales': yolact_base_config.backbone.pred_scales,
'pred_aspect_ratios': yolact_base_config.backbone.pred_aspect_ratios,
'use_pixel_scales': True,
@@ -741,7 +772,7 @@ yolact_resnet50_config = yolact_base_config.copy({
'backbone': resnet50_backbone.copy({
'selected_layers': list(range(1, 4)),
-
+
'pred_scales': yolact_base_config.backbone.pred_scales,
'pred_aspect_ratios': yolact_base_config.backbone.pred_aspect_ratios,
'use_pixel_scales': True,
@@ -753,14 +784,14 @@ yolact_resnet50_config = yolact_base_config.copy({
yolact_resnet50_pascal_config = yolact_resnet50_config.copy({
'name': None, # Will default to yolact_resnet50_pascal
-
+
# Dataset stuff
'dataset': pascal_sbd_dataset,
'num_classes': len(pascal_sbd_dataset.class_names) + 1,
'max_iter': 120000,
'lr_steps': (60000, 100000),
-
+
'backbone': yolact_resnet50_config.backbone.copy({
'pred_scales': [[32], [64], [128], [256], [512]],
'use_square_anchors': False,
@@ -774,7 +805,7 @@ yolact_plus_base_config = yolact_base_config.copy({
'backbone': resnet101_dcn_inter3_backbone.copy({
'selected_layers': list(range(1, 4)),
-
+
'pred_aspect_ratios': [ [[1, 1/2, 2]] ]*5,
'pred_scales': [[i * 2 ** (j / 3.0) for j in range(3)] for i in [24, 48, 96, 192, 384]],
'use_pixel_scales': True,
@@ -796,7 +827,7 @@ yolact_plus_resnet50_config = yolact_plus_base_config.copy({
'backbone': resnet50_dcnv2_backbone.copy({
'selected_layers': list(range(1, 4)),
-
+
'pred_aspect_ratios': [ [[1, 1/2, 2]] ]*5,
'pred_scales': [[i * 2 ** (j / 3.0) for j in range(3)] for i in [24, 48, 96, 192, 384]],
'use_pixel_scales': True,
@@ -823,4 +854,4 @@ def set_cfg(config_name:str):
def set_dataset(dataset_name:str):
""" Sets the dataset of the current config. """
cfg.dataset = eval(dataset_name)
-
+
@@ -1,2 +1 @@
from .functions import *
-from .modules import *
@@ -1,3 +1,34 @@
+# BSD 3-Clause License
+#
+# Copyright (c) 2017 xxxx
+# All rights reserved.
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# * Redistributions of source code must retain the above copyright notice, this
+# list of conditions and the following disclaimer.
+#
+# * Redistributions in binary form must reproduce the above copyright notice,
+# this list of conditions and the following disclaimer in the documentation
+# and/or other materials provided with the distribution.
+#
+# * Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+# ============================================================================
# -*- coding: utf-8 -*-
import torch
from utils import timer
@@ -1,3 +1,34 @@
+# BSD 3-Clause License
+#
+# Copyright (c) 2017 xxxx
+# All rights reserved.
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# * Redistributions of source code must retain the above copyright notice, this
+# list of conditions and the following disclaimer.
+#
+# * Redistributions in binary form must reproduce the above copyright notice,
+# this list of conditions and the following disclaimer in the documentation
+# and/or other materials provided with the distribution.
+#
+# * Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+# ============================================================================
import torch
import torch.nn.functional as F
from ..box_utils import decode, jaccard, index2d
@@ -29,7 +60,7 @@ class Detect(object):
self.use_cross_class_nms = False
self.use_fast_nms = False
- def __call__(self, predictions, net):
+ def __call__(self, predictions):
"""
Args:
loc_data: (tensor) Loc preds from loc layers
@@ -73,7 +104,7 @@ class Detect(object):
if result is not None and proto_data is not None:
result['proto'] = proto_data[batch_idx]
- out.append({'detection': result, 'net': net})
+ out.append(result)
return out
@@ -205,7 +236,7 @@ class Detect(object):
if cls_scores.size(0) == 0:
continue
- preds = torch.cat([boxes[conf_mask], cls_scores[:, None]], dim=1).cpu().numpy()
+ preds = torch.cat([boxes[conf_mask], cls_scores[:, None]], dim=1).detach().numpy()
keep = cnms(preds, iou_threshold)
keep = torch.Tensor(keep, device=boxes.device).long()
@@ -1,3 +1,34 @@
+# BSD 3-Clause License
+#
+# Copyright (c) 2017 xxxx
+# All rights reserved.
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# * Redistributions of source code must retain the above copyright notice, this
+# list of conditions and the following disclaimer.
+#
+# * Redistributions in binary form must reproduce the above copyright notice,
+# this list of conditions and the following disclaimer in the documentation
+# and/or other materials provided with the distribution.
+#
+# * Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+# ============================================================================
import torch.nn as nn
import torch.nn.functional as F
@@ -1,3 +1,34 @@
+# BSD 3-Clause License
+#
+# Copyright (c) 2017 xxxx
+# All rights reserved.
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# * Redistributions of source code must retain the above copyright notice, this
+# list of conditions and the following disclaimer.
+#
+# * Redistributions in binary form must reproduce the above copyright notice,
+# this list of conditions and the following disclaimer in the documentation
+# and/or other materials provided with the distribution.
+#
+# * Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+# ============================================================================
""" Contains functions used to sanitize and prepare the output of Yolact. """
@@ -33,8 +64,6 @@ def postprocess(det_output, w, h, batch_idx=0, interpolation_mode='bilinear',
"""
dets = det_output[batch_idx]
- net = dets['net']
- dets = dets['detection']
if dets is None:
return [torch.Tensor()] * 4 # Warning, this is 4 copies of the same thing
@@ -76,16 +105,6 @@ def postprocess(det_output, w, h, batch_idx=0, interpolation_mode='bilinear',
# Permute into the correct output shape [num_dets, proto_h, proto_w]
masks = masks.permute(2, 0, 1).contiguous()
- if cfg.use_maskiou:
- with timer.env('maskiou_net'):
- with torch.no_grad():
- maskiou_p = net.maskiou_net(masks.unsqueeze(1))
- maskiou_p = torch.gather(maskiou_p, dim=1, index=classes.unsqueeze(1)).squeeze(1)
- if cfg.rescore_mask:
- if cfg.rescore_bbox:
- scores = scores * maskiou_p
- else:
- scores = [scores, scores * maskiou_p]
# Scale masks up to the full image
masks = F.interpolate(masks.unsqueeze(0), (h, w), mode=interpolation_mode, align_corners=False).squeeze(0)
@@ -1,3 +1,34 @@
+# BSD 3-Clause License
+#
+# Copyright (c) 2017 xxxx
+# All rights reserved.
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# * Redistributions of source code must retain the above copyright notice, this
+# list of conditions and the following disclaimer.
+#
+# * Redistributions in binary form must reproduce the above copyright notice,
+# this list of conditions and the following disclaimer in the documentation
+# and/or other materials provided with the distribution.
+#
+# * Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+# ============================================================================
import torch
from torchvision import transforms
import cv2
@@ -1,3 +1,34 @@
+# BSD 3-Clause License
+#
+# Copyright (c) 2017 xxxx
+# All rights reserved.
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# * Redistributions of source code must retain the above copyright notice, this
+# list of conditions and the following disclaimer.
+#
+# * Redistributions in binary form must reproduce the above copyright notice,
+# this list of conditions and the following disclaimer in the documentation
+# and/or other materials provided with the distribution.
+#
+# * Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+# ============================================================================
import torch
import torch.nn as nn
import os
@@ -1,3 +1,34 @@
+# BSD 3-Clause License
+#
+# Copyright (c) 2017 xxxx
+# All rights reserved.
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# * Redistributions of source code must retain the above copyright notice, this
+# list of conditions and the following disclaimer.
+#
+# * Redistributions in binary form must reproduce the above copyright notice,
+# this list of conditions and the following disclaimer in the documentation
+# and/or other materials provided with the distribution.
+#
+# * Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+# ============================================================================
import time
from collections import defaultdict
@@ -1,4 +1,35 @@
-import torch, torchvision
+# BSD 3-Clause License
+#
+# Copyright (c) 2017 xxxx
+# All rights reserved.
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# * Redistributions of source code must retain the above copyright notice, this
+# list of conditions and the following disclaimer.
+#
+# * Redistributions in binary form must reproduce the above copyright notice,
+# this list of conditions and the following disclaimer in the documentation
+# and/or other materials provided with the distribution.
+#
+# * Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+# ============================================================================
+import torch
import torch.nn as nn
import torch.nn.functional as F
from torchvision.models.resnet import Bottleneck
@@ -7,22 +38,18 @@ from itertools import product
from math import sqrt
from typing import List
from collections import defaultdict
-
from data.config import cfg, mask_type
from layers import Detect
-from layers.interpolate import InterpolateModule
from backbone import construct_backbone
-
-import torch.backends.cudnn as cudnn
from utils import timer
from utils.functions import MovingAverage, make_net
# This is required for Pytorch 1.0.1 on Windows to initialize Cuda on some driver versions.
# See the bug report here: https://github.com/pytorch/pytorch/issues/17108
-torch.cuda.current_device()
+#torch.cuda.current_device()
# As of March 10, 2019, Pytorch DataParallel still doesn't support JIT Script Modules
-use_jit = torch.cuda.device_count() <= 1
+use_jit = False
if not use_jit:
print('Multiple GPUs detected! Turning off JIT.')
@@ -211,6 +238,18 @@ class PredictionModule(nn.Module):
return preds
+ @classmethod
+ def get_YOLACT_priors(cls):
+ conv_lengths = [69, 35, 18, 9, 5]
+ scales = [[24], [48], [96], [192], [384]]
+ priors_list = []
+ for index in range(0, 5):
+ cls.scales = scales[index]
+ cls.last_img_size = None
+ cls.aspect_ratios = [[1, 0.5, 2]]
+ priors_list.append(cls.make_priors(cls, conv_lengths[index], conv_lengths[index], 'cpu'))
+ return torch.cat(priors_list, -2)
+
def make_priors(self, conv_h, conv_w, device):
""" Note that priors are [x,y,width,height] where (x,y) is the center of the box. """
global prior_cache
@@ -245,7 +284,7 @@ class PredictionModule(nn.Module):
prior_data += [x, y, w, h]
- self.priors = torch.Tensor(prior_data, device=device).view(-1, 4).detach()
+ self.priors = torch.Tensor(prior_data).to(device).view(-1, 4).detach()
self.priors.requires_grad = False
self.last_img_size = (cfg._tmp_img_w, cfg._tmp_img_h)
self.last_conv_size = (conv_w, conv_h)
@@ -283,7 +322,7 @@ class FPN(ScriptModuleWrapper):
def __init__(self, in_channels):
super().__init__()
- self.lat_layers = nn.ModuleList([
+ self.lat_layers = nn.ModuleList([
nn.Conv2d(x, cfg.fpn.num_features, kernel_size=1)
for x in reversed(in_channels)
])
@@ -398,20 +437,20 @@ class Yolact(nn.Module):
def __init__(self):
super().__init__()
+ self.exportOnnx = False
+ self.backbone = construct_backbone(cfg.backbone) #backbone: resnetbackbone. backbone_modules:{list:104}. bn1:{BatchNorm2d}
- self.backbone = construct_backbone(cfg.backbone)
-
- if cfg.freeze_bn:
+ if cfg.freeze_bn: # it's true
self.freeze_bn()
# Compute mask_dim here and add it back to the config. Make sure Yolact's constructor is called early!
if cfg.mask_type == mask_type.direct:
cfg.mask_dim = cfg.mask_size**2
- elif cfg.mask_type == mask_type.lincomb:
+ elif cfg.mask_type == mask_type.lincomb: # the module will execute this branch
if cfg.mask_proto_use_grid:
self.grid = torch.Tensor(np.load(cfg.mask_proto_grid_file))
self.num_grids = self.grid.size(0)
- else:
+ else: # the module will execute this branch
self.num_grids = 0
self.proto_src = cfg.mask_proto_src
@@ -420,7 +459,7 @@ class Yolact(nn.Module):
elif cfg.fpn is not None: in_channels = cfg.fpn.num_features
else: in_channels = self.backbone.channels[self.proto_src]
in_channels += self.num_grids
-
+ #in_channels will be 256
# The include_last_relu=false here is because we might want to change it to another function
self.proto_net, cfg.mask_dim = make_net(in_channels, cfg.mask_proto_net, include_last_relu=False)
@@ -431,10 +470,10 @@ class Yolact(nn.Module):
self.selected_layers = cfg.backbone.selected_layers
src_channels = self.backbone.channels
- if cfg.use_maskiou:
+ if cfg.use_maskiou: #false
self.maskiou_net = FastMaskIoUNet()
- if cfg.fpn is not None:
+ if cfg.fpn is not None: #true
# Some hacky rewiring to accomodate the FPN
self.fpn = FPN([src_channels[i] for i in self.selected_layers])
self.selected_layers = list(range(len(self.selected_layers) + cfg.fpn.num_downsample))
@@ -447,7 +486,7 @@ class Yolact(nn.Module):
for idx, layer_idx in enumerate(self.selected_layers):
# If we're sharing prediction module weights, have every module's parent be the first one
parent = None
- if cfg.share_prediction_module and idx > 0:
+ if cfg.share_prediction_module and idx > 0: #cfg.share_prediction_module is True
parent = self.prediction_layers[0]
pred = PredictionModule(src_channels[layer_idx], src_channels[layer_idx],
@@ -458,12 +497,12 @@ class Yolact(nn.Module):
self.prediction_layers.append(pred)
# Extra parameters for the extra losses
- if cfg.use_class_existence_loss:
+ if cfg.use_class_existence_loss: #false
# This comes from the smallest layer selected
# Also note that cfg.num_classes includes background
self.class_existence_fc = nn.Linear(src_channels[-1], cfg.num_classes - 1)
- if cfg.use_semantic_segmentation_loss:
+ if cfg.use_semantic_segmentation_loss: #true
self.semantic_seg_conv = nn.Conv2d(src_channels[0], cfg.num_classes-1, kernel_size=1)
# For use in evaluation
@@ -474,9 +513,9 @@ class Yolact(nn.Module):
""" Saves the model's weights using compression because the file sizes were getting too big. """
torch.save(self.state_dict(), path)
- def load_weights(self, path):
+ def load_weights(self, path, useCuda = True):
""" Loads weights from a compressed save file. """
- state_dict = torch.load(path)
+ state_dict = torch.load(path) if useCuda else torch.load(path, map_location=torch.device('cpu'))
# For backward compatability, remove these (the new variable is called layers)
for key in list(state_dict.keys()):
@@ -673,7 +712,10 @@ class Yolact(nn.Module):
else:
pred_outs['conf'] = F.softmax(pred_outs['conf'], -1)
- return self.detect(pred_outs, self)
+ if self.exportOnnx:
+ return pred_outs
+ else:
+ return self.detect(pred_outs)