bat67 / Pytorch Fcn Easiest Demo
Licence: other
PyTorch Implementation of Fully Convolutional Networks (a very simple and easy demo).
Stars: ✭ 138
Programming Languages
python
139335 projects - #7 most used programming language
Projects that are alternatives of or similar to Pytorch Fcn Easiest Demo
Seg Mentor
TFslim based semantic segmentation models, modular&extensible boutique design
Stars: ✭ 43 (-68.84%)
Mutual labels: semantic-segmentation, fcn
Pytorch Semantic Segmentation
segmentation repo using pytorch
Stars: ✭ 75 (-45.65%)
Mutual labels: semantic-segmentation, fcn
Fcn Googlenet
GoogLeNet implementation of Fully Convolutional Networks for Semantic Segmentation in TensorFlow
Stars: ✭ 45 (-67.39%)
Mutual labels: semantic-segmentation, fcn
Espnetv2
A light-weight, power efficient, and general purpose convolutional neural network
Stars: ✭ 377 (+173.19%)
Mutual labels: cnn, semantic-segmentation
Deep Residual Unet
ResUNet, a semantic segmentation model inspired by the deep residual learning and UNet. An architecture that take advantages from both(Residual and UNet) models.
Stars: ✭ 97 (-29.71%)
Mutual labels: semantic-segmentation, fcn
Multiclass Semantic Segmentation Camvid
Tensorflow 2 implementation of complete pipeline for multiclass image semantic segmentation using UNet, SegNet and FCN32 architectures on Cambridge-driving Labeled Video Database (CamVid) dataset.
Stars: ✭ 67 (-51.45%)
Mutual labels: semantic-segmentation, fcn
DDUnet-Modified-Unet-for-WMH-with-Dense-Dilate
WMH segmentaion with unet, dilated_unet, and with ideas from denseNet
Stars: ✭ 23 (-83.33%)
Mutual labels: fcn, semantic-segmentation
Geo Deep Learning
Deep learning applied to georeferenced datasets
Stars: ✭ 91 (-34.06%)
Mutual labels: cnn, semantic-segmentation
Wb color augmenter
WB color augmenter improves the accuracy of image classification and image semantic segmentation methods by emulating different WB effects (ICCV 2019) [Python & Matlab].
Stars: ✭ 89 (-35.51%)
Mutual labels: cnn, semantic-segmentation
Piwise
Pixel-wise segmentation on VOC2012 dataset using pytorch.
Stars: ✭ 365 (+164.49%)
Mutual labels: semantic-segmentation, fcn
Segmentation
Tensorflow implementation : U-net and FCN with global convolution
Stars: ✭ 101 (-26.81%)
Mutual labels: semantic-segmentation, fcn
Edgenets
This repository contains the source code of our work on designing efficient CNNs for computer vision
Stars: ✭ 331 (+139.86%)
Mutual labels: cnn, semantic-segmentation
Pytorch Auto Drive
Segmentation models (ERFNet, ENet, DeepLab, FCN...) and Lane detection models (SCNN, SAD, PRNet, RESA, LSTR...) based on PyTorch 1.6 with mixed precision training
Stars: ✭ 32 (-76.81%)
Mutual labels: semantic-segmentation, fcn
TensorFlow-Advanced-Segmentation-Models
A Python Library for High-Level Semantic Segmentation Models based on TensorFlow and Keras with pretrained backbones.
Stars: ✭ 64 (-53.62%)
Mutual labels: fcn, semantic-segmentation
Jacinto Ai Devkit
Training & Quantization of embedded friendly Deep Learning / Machine Learning / Computer Vision models
Stars: ✭ 49 (-64.49%)
Mutual labels: cnn, semantic-segmentation
FCN-Segmentation-TensorFlow
FCN for Semantic Image Segmentation achieving 68.5 mIoU on PASCAL VOC
Stars: ✭ 34 (-75.36%)
Mutual labels: fcn, semantic-segmentation
Tools To Design Or Visualize Architecture Of Neural Network
Tools to Design or Visualize Architecture of Neural Network
Stars: ✭ 1,143 (+728.26%)
Mutual labels: cnn, semantic-segmentation
Pytorch Fcn
PyTorch Implementation of Fully Convolutional Networks. (Training code to reproduce the original result is available.)
Stars: ✭ 1,351 (+878.99%)
Mutual labels: semantic-segmentation, fcn
pytorch FCN easiest demo
不断更新中~
这个repo是在读论文Fully Convolutional Networks for Semantic Segmentation时的一个pytorch简单复现，数据集很小，是一些随机背景上的一些包的图片（所有数据集大小一共不到80M），如下图
关于此数据集详细信息，见数据集
根据论文实现了FCN32s、FCN16s、FCN8s和FCNs
部分代码参考了这个repo
使用visdom可视化，运行了20个epoch后的可视化如下图：
1.如何运行
1.1 我的运行环境
- Windows 10
- CUDA 9.x （可选）
- Anaconda 3 （numpy、os、datetime、matplotlib）
- pytorch == 0.4.1 or 1.0
- torchvision == 0.2.1
- visdom == 0.1.8.5
- OpenCV-Python == 3.4.1
1.2 具体操作
- 打开终端，输入
python -m visdom.server
- 打开另一终端，输入
python train.py
- 若没有问题可以打开浏览器输入
http://localhost:8097/来使用visdom可视化
1.3 训练细节
2. 数据集
- training data来自这里，ground-truth来自这里。
- 链接中提供的图片中，部分ground-truth的有误，而且部分有ground-truth的图片没有对应training data的图片，将这些有错误的图片分别剔除，重新编号排序之后剩余533张图片。
- 之后我随机选取了67张图片旋转180度，一共在training data和ground-truth分别凑够600张图片（0.jpg ~ 599.jpg）。
3. 可视化
- train prediction：训练时模型的输出
- label：ground-truth
- test prediction：预测时模型的输出（每次训练都会预测，但预测数据不参与训练与backprop）
- train iter loss：训练时每一批（batch）的loss情况
- test iter loss：测试时每一批（batch）的loss情况
4. 包含文件
4.1 train.py
- 训练网络与可视化
- 主函数
4.2 FCN.py
- FCN32s、FCN16s、FCN8s、FCNs网络定义
- VGGNet网络定义、VGG不同种类网络参数、构建VGG网络的函数
4.3 BagData.py
- 定义方便PyTorch读取数据的Dataset和DataLoader
- 定义数据的变换transform
4.4 onehot.py
- 图片的onehot编码
Note that the project description data, including the texts, logos, images, and/or trademarks,
for each open source project belongs to its rightful owner.
If you wish to add or remove any projects, please contact us at [email protected].