大年初一我居然在更博客。今年过年由于病毒横行，没有串门没有聚餐，整个人闲的没事干。。。医生真是不容易，忙得团团转还有生命危险，新希望他们平安。

本篇不属于初级教程。如果完全看不懂请自行谷歌或搜索作者博客。

deeplab官方提供了多种backbone，通过train.py中传递参数，

--model_variant="resnet_v1_101_beta" \

可以更改backbone。（resnet_v1_{50,101}_beta: We modify the original ResNet-101 [10], similar to PSPNet [11] by replacing the first 7x7 convolution with three 3x3 convolutions. See resnet_v1_beta.py for more details.）

这个是官方对beta的说法。就是改了一个卷积核的尺寸。改小了，并且用了三个。

一共有如下可选：

# A map from feature extractor name to the network name scope used in the
# ImageNet pretrained versions of these models.
name_scope = {'mobilenet_v2': 'MobilenetV2','resnet_v1_50': 'resnet_v1_50','resnet_v1_50_beta': 'resnet_v1_50','resnet_v1_101': 'resnet_v1_101','resnet_v1_101_beta': 'resnet_v1_101','xception_41': 'xception_41','xception_65': 'xception_65','xception_71': 'xception_71','nas_pnasnet': 'pnasnet','nas_hnasnet': 'hnasnet',
}

当然backbone更改后，网络训练的参数比如decay什么的也有所不同。在feature_extractor.py中307行开始就是在改参数，举个例子：

  if 'resnet' in model_variant:arg_scope = arg_scopes_map[model_variant](weight_decay=weight_decay,batch_norm_decay=0.95,batch_norm_epsilon=1e-5,batch_norm_scale=True)

并且在train.py中：

# For weight_decay, use 0.00004 for MobileNet-V2 or Xcpetion model variants.
# Use 0.0001 for ResNet model variants.
flags.DEFINE_float('weight_decay', 0.00004,'The value of the weight decay for training.')

所以，在我自己的bash文件中，我也要改

--weight_decay=0.0001 \

于是我完整的bash文件就是：

python "${WORK_DIR}"/train.py \--logtostderr \--train_split="train" \--model_variant="resnet_v1_50_beta" \--weight_decay=0.0001 \--atrous_rates=6 \--atrous_rates=12 \--atrous_rates=18 \--output_stride=16 \--decoder_output_stride=4 \--train_crop_size="513,513" \--train_batch_size=12 \--num_clones=6 \--training_number_of_steps=30000 \--fine_tune_batch_norm=True \--train_logdir="${TRAIN_LOGDIR}" \--dataset_dir="${PASCAL_DATASET}" \--tf_initial_checkpoint="${RES_WEIGHT}" \--base_learning_rate=0.007 \

说完了命令再说一下权重，官方给了resnet101，以及resnet50在imagenet，（https://github.com/tensorflow/models/blob/master/research/deeplab/g3doc/model_zoo.md）resnet预训练的权重。从官网下载后，加载的过程中，我发现，如果使用

--model_variant="resnet_v1_101" \

会出现加载错误。网络结构中在bottleneck上的参数设置，与checkpoint训练的网络结构不一样。同时，resnet在论文中提及的时候，作者说自己改过了。所以，这里大年初一更博客的笔者推测，beta版本才是真正的backbone。由于谷歌上不去，不想用镜像，所以笔者使用的是beta。使用后权重加载成功，并且有如下提示：


INFO:tensorflow:Initializing model from path: /home/DATA/liutian/tmp/tfdeeplab/resnet/model.ckpt
WARNING:tensorflow:Checkpoint is missing variable [image_pooling/weights]
WARNING:tensorflow:Checkpoint is missing variable [image_pooling/BatchNorm/gamma]
WARNING:tensorflow:Checkpoint is missing variable [image_pooling/BatchNorm/beta]
WARNING:tensorflow:Checkpoint is missing variable [image_pooling/BatchNorm/moving_mean]
WARNING:tensorflow:Checkpoint is missing variable [image_pooling/BatchNorm/moving_variance]
WARNING:tensorflow:Checkpoint is missing variable [aspp0/weights]
WARNING:tensorflow:Checkpoint is missing variable [aspp0/BatchNorm/gamma]
WARNING:tensorflow:Checkpoint is missing variable [aspp0/BatchNorm/beta]
WARNING:tensorflow:Checkpoint is missing variable [aspp0/BatchNorm/moving_mean]
WARNING:tensorflow:Checkpoint is missing variable [aspp0/BatchNorm/moving_variance]
WARNING:tensorflow:Checkpoint is missing variable [aspp1_depthwise/depthwise_weights]
WARNING:tensorflow:Checkpoint is missing variable [aspp1_depthwise/BatchNorm/gamma]
WARNING:tensorflow:Checkpoint is missing variable [aspp1_depthwise/BatchNorm/beta]
WARNING:tensorflow:Checkpoint is missing variable [aspp1_depthwise/BatchNorm/moving_mean]
WARNING:tensorflow:Checkpoint is missing variable [aspp1_depthwise/BatchNorm/moving_variance]
WARNING:tensorflow:Checkpoint is missing variable [aspp1_pointwise/weights]
WARNING:tensorflow:Checkpoint is missing variable [aspp1_pointwise/BatchNorm/gamma]
WARNING:tensorflow:Checkpoint is missing variable [aspp1_pointwise/BatchNorm/beta]
WARNING:tensorflow:Checkpoint is missing variable [aspp1_pointwise/BatchNorm/moving_mean]
WARNING:tensorflow:Checkpoint is missing variable [aspp1_pointwise/BatchNorm/moving_variance]
WARNING:tensorflow:Checkpoint is missing variable [aspp2_depthwise/depthwise_weights]
WARNING:tensorflow:Checkpoint is missing variable [aspp2_depthwise/BatchNorm/gamma]
WARNING:tensorflow:Checkpoint is missing variable [aspp2_depthwise/BatchNorm/beta]
WARNING:tensorflow:Checkpoint is missing variable [aspp2_depthwise/BatchNorm/moving_mean]
WARNING:tensorflow:Checkpoint is missing variable [aspp2_depthwise/BatchNorm/moving_variance]
WARNING:tensorflow:Checkpoint is missing variable [aspp2_pointwise/weights]
WARNING:tensorflow:Checkpoint is missing variable [aspp2_pointwise/BatchNorm/gamma]
WARNING:tensorflow:Checkpoint is missing variable [aspp2_pointwise/BatchNorm/beta]
WARNING:tensorflow:Checkpoint is missing variable [aspp2_pointwise/BatchNorm/moving_mean]
WARNING:tensorflow:Checkpoint is missing variable [aspp2_pointwise/BatchNorm/moving_variance]
WARNING:tensorflow:Checkpoint is missing variable [aspp3_depthwise/depthwise_weights]
WARNING:tensorflow:Checkpoint is missing variable [aspp3_depthwise/BatchNorm/gamma]
WARNING:tensorflow:Checkpoint is missing variable [aspp3_depthwise/BatchNorm/beta]
WARNING:tensorflow:Checkpoint is missing variable [aspp3_depthwise/BatchNorm/moving_mean]
WARNING:tensorflow:Checkpoint is missing variable [aspp3_depthwise/BatchNorm/moving_variance]
WARNING:tensorflow:Checkpoint is missing variable [aspp3_pointwise/weights]
WARNING:tensorflow:Checkpoint is missing variable [aspp3_pointwise/BatchNorm/gamma]
WARNING:tensorflow:Checkpoint is missing variable [aspp3_pointwise/BatchNorm/beta]
WARNING:tensorflow:Checkpoint is missing variable [aspp3_pointwise/BatchNorm/moving_mean]
WARNING:tensorflow:Checkpoint is missing variable [aspp3_pointwise/BatchNorm/moving_variance]
WARNING:tensorflow:Checkpoint is missing variable [concat_projection/weights]
WARNING:tensorflow:Checkpoint is missing variable [concat_projection/BatchNorm/gamma]
WARNING:tensorflow:Checkpoint is missing variable [concat_projection/BatchNorm/beta]
WARNING:tensorflow:Checkpoint is missing variable [concat_projection/BatchNorm/moving_mean]
WARNING:tensorflow:Checkpoint is missing variable [concat_projection/BatchNorm/moving_variance]
WARNING:tensorflow:Checkpoint is missing variable [decoder/feature_projection0/weights]
WARNING:tensorflow:Checkpoint is missing variable [decoder/feature_projection0/BatchNorm/gamma]
WARNING:tensorflow:Checkpoint is missing variable [decoder/feature_projection0/BatchNorm/beta]
WARNING:tensorflow:Checkpoint is missing variable [decoder/feature_projection0/BatchNorm/moving_mean]
WARNING:tensorflow:Checkpoint is missing variable [decoder/feature_projection0/BatchNorm/moving_variance]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv0_depthwise/depthwise_weights]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv0_depthwise/BatchNorm/gamma]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv0_depthwise/BatchNorm/beta]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv0_depthwise/BatchNorm/moving_mean]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv0_depthwise/BatchNorm/moving_variance]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv0_pointwise/weights]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv0_pointwise/BatchNorm/gamma]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv0_pointwise/BatchNorm/beta]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv0_pointwise/BatchNorm/moving_mean]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv0_pointwise/BatchNorm/moving_variance]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv1_depthwise/depthwise_weights]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv1_depthwise/BatchNorm/gamma]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv1_depthwise/BatchNorm/beta]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv1_depthwise/BatchNorm/moving_mean]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv1_depthwise/BatchNorm/moving_variance]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv1_pointwise/weights]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv1_pointwise/BatchNorm/gamma]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv1_pointwise/BatchNorm/beta]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv1_pointwise/BatchNorm/moving_mean]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv1_pointwise/BatchNorm/moving_variance]
WARNING:tensorflow:Checkpoint is missing variable [logits/semantic/weights]
WARNING:tensorflow:Checkpoint is missing variable [logits/semantic/biases]
WARNING:tensorflow:Checkpoint is missing variable [image_pooling/weights/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [image_pooling/BatchNorm/gamma/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [image_pooling/BatchNorm/beta/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [aspp0/weights/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [aspp0/BatchNorm/gamma/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [aspp0/BatchNorm/beta/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [aspp1_depthwise/depthwise_weights/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [aspp1_depthwise/BatchNorm/gamma/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [aspp1_depthwise/BatchNorm/beta/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [aspp1_pointwise/weights/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [aspp1_pointwise/BatchNorm/gamma/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [aspp1_pointwise/BatchNorm/beta/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [aspp2_depthwise/depthwise_weights/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [aspp2_depthwise/BatchNorm/gamma/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [aspp2_depthwise/BatchNorm/beta/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [aspp2_pointwise/weights/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [aspp2_pointwise/BatchNorm/gamma/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [aspp2_pointwise/BatchNorm/beta/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [aspp3_depthwise/depthwise_weights/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [aspp3_depthwise/BatchNorm/gamma/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [aspp3_depthwise/BatchNorm/beta/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [aspp3_pointwise/weights/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [aspp3_pointwise/BatchNorm/gamma/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [aspp3_pointwise/BatchNorm/beta/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [concat_projection/weights/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [concat_projection/BatchNorm/gamma/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [concat_projection/BatchNorm/beta/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [decoder/feature_projection0/weights/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [decoder/feature_projection0/BatchNorm/gamma/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [decoder/feature_projection0/BatchNorm/beta/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv0_depthwise/depthwise_weights/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv0_depthwise/BatchNorm/gamma/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv0_depthwise/BatchNorm/beta/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv0_pointwise/weights/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv0_pointwise/BatchNorm/gamma/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv0_pointwise/BatchNorm/beta/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv1_depthwise/depthwise_weights/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv1_depthwise/BatchNorm/gamma/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv1_depthwise/BatchNorm/beta/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv1_pointwise/weights/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv1_pointwise/BatchNorm/gamma/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [decoder/decoder_conv1_pointwise/BatchNorm/beta/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [logits/semantic/weights/Momentum]
WARNING:tensorflow:Checkpoint is missing variable [logits/semantic/biases/Momentum]
INFO:tensorflow:Create CheckpointSaverHook.

我认为这个说明checkpoint少了decode与aspp的网络。而beta是有的。因为在代码中，aspp是否使用是通过参数空置的（model.py 397行：

model_options.aspp_with_batch_norm

），decode是否使用也是通过参数控制的（

decoder_output_stride如果不给参数就直接删除整个decoder。

），modelvariant只会影响encoder阶段，也就是feature_extractor.

那么我的确认方式是1，debug2，看tensorboard

ValueError: Total size of new array must be unchanged for resnet_v1_101/block1/unit_1/bottleneck_v1/shortcut/weights lh_shape: [(1, 1, 128, 256)], rh_shape: [(1, 1, 64, 256)]

之所以废这么多话是想说，复现可能会有一定问题，因为你需要先用coco预训练，再用voc2012 trainaug set预训练，得到的权重才可以和论文比。

我是没用coco。

更-------------------------

这个病毒越演愈烈，正好在北京直面过非典，实际上非典时期结束有一部分是因为天气变热，病毒传播困难。再加上隔离严格。所以武汉肺炎终究会过去。就是医护人员在湖北人手不足，新闻上全家感染的例子不在少数。致死率没有非典严重，大多数是并发症。但是传染的速度真的是太快了。虽然不能恐慌，但是也要严肃对待。未感染的需要提高免疫力，摄取维生素c，注意保暖，不要感冒发烧，以免给医疗系统增加压力。

这里贴的是resnet101在voc的结果，

这个贴的是xception，可以看到，这里采用了coco以及JFT两个数据集预训练作为变量。而resnet则没有，所以很有可能是没用coco，只用了imagenet。

所以我的总的bash代码是这样的。

#!/bin/bash
# Copyright 2018 The TensorFlow Authors All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ==============================================================================
#
# This script is used to run local test on PASCAL VOC 2012. Users could also
# modify from this script for their use case.
#
# Usage:
#   # From the tensorflow/models/research/deeplab directory.
#   sh ./local_test.sh
#
## Exit immediately if a command exits with a non-zero status.
set -e# Move one-level up to tensorflow/models/research directory.
cd ..# Update PYTHONPATH.
export PYTHONPATH=$PYTHONPATH:`pwd`:`pwd`/slim# Set up the working environment.
CURRENT_DIR=$(pwd)
WORK_DIR="${CURRENT_DIR}/deeplab"
RES_DIR="${CURRENT_DIR}/resnet"
RES_WEIGHT="${CURRENT_DIR}/resnet/model.ckpt"
#export CUDA_VISIBLE_DEVICES=3
# Run model_test first to make sure the PYTHONPATH is correctly set.
#python "${WORK_DIR}"/model_test.py -v# Go to datasets folder and download PASCAL VOC 2012 segmentation dataset.
DATASET_DIR="datasets"
cd "${WORK_DIR}/${DATASET_DIR}"
#sh download_and_convert_voc2012.sh# Go back to original directory.
cd "${CURRENT_DIR}"# Set up the working directories.
PASCAL_FOLDER="pascal_voc_seg"
EXP_FOLDER="result/offres_freze"
INIT_FOLDER="${WORK_DIR}/${DATASET_DIR}/${PASCAL_FOLDER}/init_models"
COCO_PRE="${WORK_DIR}/${DATASET_DIR}/${PASCAL_FOLDER}/coco_pretrain"
TRAIN_LOGDIR="${WORK_DIR}/${DATASET_DIR}/${PASCAL_FOLDER}/${EXP_FOLDER}/train"
VOC_LOGDIR="${WORK_DIR}/${DATASET_DIR}/${PASCAL_FOLDER}/${EXP_FOLDER}/train_voc"
EVAL_LOGDIR="${WORK_DIR}/${DATASET_DIR}/${PASCAL_FOLDER}/${EXP_FOLDER}/eval"
VIS_LOGDIR="${WORK_DIR}/${DATASET_DIR}/${PASCAL_FOLDER}/${EXP_FOLDER}/vis"
EXPORT_DIR="${WORK_DIR}/${DATASET_DIR}/${PASCAL_FOLDER}/${EXP_FOLDER}/export"
mkdir -p "${INIT_FOLDER}"
mkdir -p "${TRAIN_LOGDIR}"
mkdir -p "${EVAL_LOGDIR}"
mkdir -p "${VIS_LOGDIR}"
mkdir -p "${EXPORT_DIR}"# Copy locally the trained checkpoint as the initial checkpoint.
TF_INIT_ROOT="http://download.tensorflow.org/models"
TF_INIT_CKPT="deeplabv3_pascal_train_aug_2018_01_04.tar.gz"
cd "${INIT_FOLDER}"
#wget -nd -c "${TF_INIT_ROOT}/${TF_INIT_CKPT}"cd "${CURRENT_DIR}"PASCAL_DATASET="${WORK_DIR}/${DATASET_DIR}/${PASCAL_FOLDER}/tfrecord"# Train 10 iterations.
NUM_ITERATIONS=20
echo ${PASCAL_DATASET}python "${WORK_DIR}"/train.py \--logtostderr \--train_split="train_aug" \--model_variant="resnet_v1_101_beta" \--weight_decay=0.001 \--atrous_rates=6 \--atrous_rates=12 \--atrous_rates=18 \--output_stride=16 \--decoder_output_stride=4 \--train_crop_size="513,513" \--train_batch_size=32 \--num_clones=8 \--training_number_of_steps=30000 \--fine_tune_batch_norm=True \--train_logdir="${TRAIN_LOGDIR}" \--dataset_dir="${PASCAL_DATASET}" \--tf_initial_checkpoint="${RES_WEIGHT}" \python "${WORK_DIR}"/train.py \--logtostderr \--train_split="train" \--model_variant="resnet_v1_101_beta" \--weight_decay=0.0001 \--atrous_rates=6 \--atrous_rates=12 \--atrous_rates=18 \--output_stride=16 \--decoder_output_stride=4 \--train_crop_size="513,513" \--train_batch_size=32 \--num_clones=8 \--training_number_of_steps=30000 \--fine_tune_batch_norm=True \--train_logdir="${VOC_LOGDIR}" \--dataset_dir="${PASCAL_DATASET}" \--tf_initial_checkpoint="${TRAIN_LOGDIR}/model.ckpt-30000" \

论文较真：

由于官方没写resnet作为backbone怎么训练的，所以这里放一个同样用deeplabv3+，resnet，cvpr的文章：

Decoders Matter for Semantic Segmentation:Data-Dependent Decoding Enables Flexible Feature Aggregation

在val的pascal voc数据集大概是72,73的准确率：

官方代码 Deeplab v3+ resnet101 做backbone相关推荐

  1. 附代码 Deeplab V3

    Rethinking Atrous Convolution for Semantic Image Segmentation 论文解读 参考链接:https://zhuanlan.zhihu.com/p ...

  2. Deeplab v3+的结构代码简要分析--Pytorch 版

    转载自:https://www.cnblogs.com/ywheunji/p/10479019.html.侵删 添加了解码模块来重构精确的图像物体边界.对比如图 deeplab v3+采用了与deep ...

  3. deeplab v3+---Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation

    一.摘要: 1.spp是什么? 问题:分割 我们提出了什么: 1.deeplab v3+ 在deeplab v3的基础上加了一个简单的decoder模块来改善分割结果,尤其是对于边界区域 2.我们采用 ...

  4. DeepLab V3学习笔记

    DeepLab V3 遇到的问题和解决方法 相关工作 DeepLab V3中的两种模型结构 cascaded model ASPP model 相对于DeepLab V2的优化 Multi-grid ...

  5. 一文理解DeepLab V1到DeepLab V3+（超系统、超详细）

    文章目录 预备知识 空洞卷积 1.DeepLab V1 1.1 基于VGG模型 1.2 总体架构 1.2.1 Fully Connected CRF(条件随机场) 1.3 DeepLab V1模型实验 ...

  6. Deeplab V3+训练自己数据集全过程

    Deeplab V3+训练过程 0. 参考链接 1. 制作数据集 1.1. 制作json文件. 1.2. 批量转换 json--png 1.3. 数据集目录制作 1.4. 生成tfrecord 2. ...

  7. deeplab v3+论文精读

    论文地址:https://arxiv.org/pdf/1802.02611.pdf tensorflow代码:https: //github.com/tensorflow/models/tree/ma ...

  8. DeepLab v3+为啥可以封神？（论文讲解含超详细注解+中英文对照+配图）

    开始前,博主请求大家一定要看注解,博主的努力全在注解里,有帮助的记得一键三连呀! Encoder-Decoder with Atrous Separable Convolution for Seman ...

  9. deeplab v3+ 源码详解

    训练模型:         下载好voc数据集,并传入所需的参数即可进行训练. 参数配置: """ 训练: --model deeplabv3plus_mobilenet ...

  10. 关于「Xception」和「DeepLab V3+」的那些事

    作者丨崔权 学校丨早稻田大学硕士生 研究方向丨深度学习,计算机视觉 知乎专栏丨サイ桑的炼丹炉 前言 最近读了 Xception [1] 和 DeepLab V3+ [2] 的论文,觉得有必要总结一下这 ...

最新文章

  1. 微信拦截网站怎么办 微信屏蔽网址如何正常打开
  2. FD.io/VPP — VPP Agent — Overview
  3. Buffer Pool--内存总结2
  4. linux centos删除安装的包,CentOS yum认为已删除的软件包仍在安装中
  5. error LNK2001: unresolved external symbol QtCored.lib using staic Qt lib
  6. 来入门一下kotlin吧
  7. 小程序【笔记001】框架和配置文件
  8. 嵌入式linux的学习笔记-pipe管道(二)
  9. mysql数据库数字类型_Mysql 数据库数值类型详解
  10. 用ghost备份和还原Linux系统(一)
  11. python支持esc退出_听说你不会用mac电脑装Python？看了这篇文章，秒懂！
  12. python需要的项目依赖包的使用
  13. java实现生命游戏
  14. [ZT]狄马:老子与美国南北战争
  15. 男主龙失忆java_男主失忆的小说推荐：我忘了全世界，却记得对你的爱，一生不变...
  16. 电子或通信领域当前的技术及其社会需求调查报告
  17. 基于linux的贪吃蛇游戏设计_贪吃蛇还能这么玩？绝对是你从未体验过的全新版本（上）...
  18. 为什么使用C#开发软件的公司和程序员都很少？
  19. 如何解决Mac与iPhone之间handoff连接问题
  20. 第十章 进程间的通信 之 Java/Android多线程开发（二）

热门文章

  1. LG android TV 蓝牙,LG TV Plus
  2. 版本控制工具-Git
  3. java基础-面向对象
  4. 安全合规/等级保护--13--我们通过了等级保护三级认证
  5. 员工请假审批系统 php,php073企业考勤请假系统
  6. 维修电工技师、高级技师技能实训考核装置
  7. IDEA主题分享与代码颜色设置
  8. linux nginx rtmp 直播,nginx+rtmp简单直播
  9. php gd jpeg,怎么解决GD库不支持JPEG
  10. php环境配置PHPWAMP