hetong007 / Gluon Fashionai Attributes
Stars: ✭ 176
Labels
Projects that are alternatives of or similar to Gluon Fashionai Attributes
Prml
Python implementations (on jupyter notebook) of algorithms described in the book "PRML"
Stars: ✭ 174 (-1.14%)
Mutual labels: jupyter-notebook
Intro Spacy Nlp
An introduction to using spaCy for NLP and machine learning
Stars: ✭ 175 (-0.57%)
Mutual labels: jupyter-notebook
Deep Algotrading
A resource for learning about deep learning techniques from regression to LSTM and Reinforcement Learning using financial data and the fitness functions of algorithmic trading
Stars: ✭ 173 (-1.7%)
Mutual labels: jupyter-notebook
Poems generator keras
唐诗,藏头诗,按需自动生成古诗,基于Keras、LSTM-RNN。文档齐全。
Stars: ✭ 175 (-0.57%)
Mutual labels: jupyter-notebook
Recommendation
Recommendation System using ML and DL
Stars: ✭ 174 (-1.14%)
Mutual labels: jupyter-notebook
Machine Learning And Reinforcement Learning In Finance
Machine Learning and Reinforcement Learning in Finance New York University Tandon School of Engineering
Stars: ✭ 173 (-1.7%)
Mutual labels: jupyter-notebook
Attentionn
All about attention in neural networks. Soft attention, attention maps, local and global attention and multi-head attention.
Stars: ✭ 175 (-0.57%)
Mutual labels: jupyter-notebook
Web Database Analytics
Web scrapping and related analytics using Python tools
Stars: ✭ 175 (-0.57%)
Mutual labels: jupyter-notebook
Deep Learning With Keras Notebooks
Jupyter notebooks for using & learning Keras
Stars: ✭ 2,077 (+1080.11%)
Mutual labels: jupyter-notebook
Astropy Tutorials
Tutorials for the Astropy Project
Stars: ✭ 174 (-1.14%)
Mutual labels: jupyter-notebook
Muffin Cupcake
classifying muffin and cupcake recipes using support vector machines
Stars: ✭ 175 (-0.57%)
Mutual labels: jupyter-notebook
Tech
Documentation of all collective action from tech workers.
Stars: ✭ 176 (+0%)
Mutual labels: jupyter-notebook
Analytical Tutorials
Tutorials for writing analytical scripts
Stars: ✭ 175 (-0.57%)
Mutual labels: jupyter-notebook
Python Machine Learning Book 3rd Edition
The "Python Machine Learning (3rd edition)" book code repository
Stars: ✭ 2,883 (+1538.07%)
Mutual labels: jupyter-notebook
Gluon-FashionAI-Attributes
这是为阿里天池竞赛——服饰属性标签识别提供的gluon
教程与benchmark代码。
gluon教程会从零开始一步步讲解,带你上手此次竞赛,同时也能帮助理解benchmark代码。
当前benchmark代码
- 在初赛数据集上能达到大约0.95的mAP与0.84的准确率,同时还有很大的提升空间。
- 在4块Tesla V100以及32核CPU的 AWS p3.8xlarge 机器上的总运行时间约为2.5小时。
重现benchmark步骤:
- 前往比赛官网,登录并注册参赛后可在左边“赛题与数据”标签内下载数据。之后将数据解压到
data/
文件夹中。解压后的目录结构应该如下所示:
Gluon-FashionAI-Attributes
├── benchmark.sh
├── data
│ ├── base
│ ├── rank
│ └── web
├── FashionAI-Attributes-Skirt.ipynb
├── prepare_data.py
├── README.md
└── train_task.py
- 参考FashionAI-Attributes-Skirt.ipynb中的环境配置一节配置教程
- 根据具体运行环境设置
benchmark.sh
中的变量
-
num_gpus
,即GPU的个数,设置为0即为只用CPU训练。 -
num_workers
,即用来处理数据的进程个数,建议设置为CPU核心个数。
- 运行
bash benchmark.sh
,这个脚本会自动准备数据,针对每个任务训练模型并预测,以及最后的合并预测。 - 运行结束后,将
submission/submission.csv
压缩成zip
格式并通过官网左侧的“提交结果”标签页提交。
保存/读取模型文件
在参加比赛时,选手们常常会训练多个模型,有时也需要保存模型留作以后使用。
如果要保存模型,可以在训练过程中使用
finetune_net.save_params('filename.params')
命令来实现。在读取时,可以通过
net = gluon.model_zoo.vision.get_model(model_name)
with net.name_scope():
net.output = nn.Dense(task_num_class)
net.load_params('filename.params', ctx=mx.gpu())
来导入模型文件。
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].