jaeoh2 / Road Lane Instance Segmentation Pytorch
Licence: mit
tuSimple dataset road lane instance segmentation with PyTorch, ROS, ENet, SegNet and Discriminative Loss.
Stars: ✭ 59
Labels
Projects that are alternatives of or similar to Road Lane Instance Segmentation Pytorch
One Pixel Attack Keras
Keras implementation of "One pixel attack for fooling deep neural networks" using differential evolution on Cifar10 and ImageNet
Stars: ✭ 1,097 (+1759.32%)
Mutual labels: jupyter-notebook
Apparel Styles
Deep Learning to Describe Clothing by Semantic Attributes
Stars: ✭ 58 (-1.69%)
Mutual labels: jupyter-notebook
Ia4business
Curso de Inteligencia Artificial aplicada a Negocios y Empresas
Stars: ✭ 59 (+0%)
Mutual labels: jupyter-notebook
C3d
C3D is a modified version of BVLC caffe to support 3D ConvNets.
Stars: ✭ 1,097 (+1759.32%)
Mutual labels: jupyter-notebook
Fraud Detection
Credit Card Fraud Detection using ML: IEEE style paper + Jupyter Notebook
Stars: ✭ 58 (-1.69%)
Mutual labels: jupyter-notebook
Principles Of Data Science
Principles of Data Science by Packt
Stars: ✭ 58 (-1.69%)
Mutual labels: jupyter-notebook
Jupyter Rise
Auto-launching the RISE plugin for Binder presentations.
Stars: ✭ 58 (-1.69%)
Mutual labels: jupyter-notebook
Deej A.i.
Create automatic playlists by using Deep Learning to *listen* to the music
Stars: ✭ 57 (-3.39%)
Mutual labels: jupyter-notebook
Cloudml Magic
Jupyter Notebook Magics for Google Cloud ML Engine
Stars: ✭ 58 (-1.69%)
Mutual labels: jupyter-notebook
Mind
2020 MIND news recomendation first place solution
Stars: ✭ 59 (+0%)
Mutual labels: jupyter-notebook
A journey into math of ml
汉语自然语言处理视频教程-开源学习资料
Stars: ✭ 1,094 (+1754.24%)
Mutual labels: jupyter-notebook
Big Mac Data
Data and methodology for the Big Mac index
Stars: ✭ 1,095 (+1755.93%)
Mutual labels: jupyter-notebook
College Majors Visualisation
Interactive Data Visualisation of U.S College Majors
Stars: ✭ 59 (+0%)
Mutual labels: jupyter-notebook
Pyspark Examples
Code examples on Apache Spark using python
Stars: ✭ 58 (-1.69%)
Mutual labels: jupyter-notebook
Attentional Interfaces
🔍 Attentional interfaces in TensorFlow.
Stars: ✭ 58 (-1.69%)
Mutual labels: jupyter-notebook
Ssd keras
Port of Single Shot MultiBox Detector to Keras
Stars: ✭ 1,101 (+1766.1%)
Mutual labels: jupyter-notebook
Pytorchdiscreteflows
Discrete Normalizing Flows implemented in PyTorch
Stars: ✭ 59 (+0%)
Mutual labels: jupyter-notebook
Road-Lane-Instance-Segmentation-PyTorch
Road lane instance segmentation with PyTorch.
- SegNet, ENet with discriminative loss.
- Lane clustered with DBSCAN.
- Trained from tuSimple dataset.
- ROS(Robot Operating System) inference node (20Hz).
ENet result
SegNet result
ROS
$ python2 ros_lane_detect.py --model-path model_best_enet.pth
Train
$ mkdir logs
$ tensorboard --logdir=logs/ &
$ python3 train.py --train-path /tuSimple/train_set/ --epoch 100 --batch-size 16 --lr 0.0001 --img-size 224 224
Dataset
Downloads: tuSimple dataset
Load Dataset
train_path = '/data/tuSimple/train_set/'
train_dataset = tuSimpleDataset(train_path, size=SIZE)
train_dataloader = torch.utils.data.DataLoader(dataset=train_dataset, batch_size=BATCH_SIZE, shuffle=True, num_workers=16)
Model
ENet summary
----------------------------------------------------------------
Total params: 686,058
Trainable params: 686,058
Non-trainable params: 0
----------------------------------------------------------------
Input size (MB): 0.57
Forward/backward pass size (MB): 153326.17
Params size (MB): 2.62
Estimated Total Size (MB): 153329.36
----------------------------------------------------------------
SegNet summary
----------------------------------------------------------------
Total params: 29,447,047
Trainable params: 29,447,047
Non-trainable params: 0
----------------------------------------------------------------
Input size (MB): 0.57
Forward/backward pass size (MB): 688.68
Params size (MB): 112.33
Estimated Total Size (MB): 801.59
----------------------------------------------------------------
References
https://github.com/nyoki-mtl/pytorch-discriminative-loss
Paper: Semantic Instance Segmentation with a Discriminative Loss Function
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].