This project provides a framework based on docker and aims to expedite the human activity classification training process. Thus, three separate environments are provided:
- Pre-processing environment.
- Training environment.
- Inference environment
While the training environment has a more general use, providing a generic tool to solve a vast amount of problems, the pre-processing environment has its focus on pre-processing human activity datasets (measured in a quaternion form) to solve the already mentioned "human activity classification problem".
Meanwhile, the inference environment serves a development framework to deploy a flask rest API. This API loads the desired neural network model and is able to answer prediction requests. This API is also focused on the human activity classification problem, but can be easily tuned for a more generic purpose. To know more about flask you can visit the official flask webpage.
Pre-requirements:
- Docker v17
- GNU Make
The following instruction launches both environments:
# Launch the development environment.
make develenv-upAlso a make help utility is available to the developer.
For the reference there is a generic view of the architecture:
As you'd have noticed, the inference environment has little to do with the previous architecture and can be treated as an individual component. Ideally, this component will be the only one deployed in a production environment.
More information can be found here: data structure documentation
# Enter the pre-processing environment
make preprocess-shThe guide for this environment can be found here: pre-process environment documentation
# Enter the training environment
make train-shThe guide for this environment can be found here: train environment documentation
# Enter the pre-processing environment
make inference-shAlthough, the most useful command in this environment may be:
# Display container logs
docker logs -f framework_inferencer_1The guide for this environment can be found here: inference environment documentation
This project is the final assignment for Gonzalo Pardo Villalibre. The aim will be to detect which activity is a certain subject performing, minimizing the number of sensors needed. Therefore the student will take advantage of the use of NN (neural networks) from different types such as CNN (convolutional networks) or RNN (recurrent networks) such LSTM.
On this journey the developer decided to not only solve the concrete problem, but also to create a reusable framework making the process easier for future investigators.
More info about specific problem can be found here: more info
You can contact the creator via e-mail at: gonzalopmb@gmail.com
MIT License
Copyright (c) 2021 Gonzalo Pardo Villalibre
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.