Boiler plate for Fast API

Experimenting with deployments on the model

Contains:

• hello world, located at hello_world.py
• restful api, located at restful_api.py, source from chuck's repository
• main (predict model api), located at main.py

How to run

1. Install the necessary libraries located at requirements.txt by using pip install -r requirements.txt
2. Run the file with uvicorn [filename]:app --reload
3. Experiment with it or open the docs

Examples:

1. To run the hello world app

uvicorn hello_world:app --reload

2. To run the restful api app

uvicorn restful_api:app --reload

You will obtain an output similar to this in the console:

INFO:     Will watch for changes in these directories: ['E:\\Projects\\try-deploying-model']
INFO:     Uvicorn running on http://127.0.0.1:8000 (Press CTRL+C to quit)
INFO:     Started reloader process [7640] using WatchFiles
INFO:     Started server process [16504]
INFO:     Waiting for application startup.
INFO:     Application startup complete.

Then you can open the browser and receive an output:

{
  "items": {
    "0": {
      "name": "Apple",
      "price": 9.99,
      "count": 20,
      "id": 0,
      "category": "fruits"
    },
    "1": {
      "name": "Kale",
      "price": 5.99,
      "count": 20,
      "id": 1,
      "category": "vegetables"
    },
    "2": {
      "name": "Orange",
      "price": 1.99,
      "count": 100,
      "id": 2,
      "category": "fruits"
    }
  }
}

3. To run the main.py sky classification model

Note that you need to have model.h5 to run this api

You can get it through this link or run the download.py file

Note that you should name your model with model.h5 or simply change the model path in the script

uvicorn main:app --reload

You will obtain an output similar to this in the console:

INFO:     Will watch for changes in these directories: ['E:\\Projects\\try-deploying-model']
INFO:     Uvicorn running on http://127.0.0.1:8000 (Press CTRL+C to quit)
INFO:     Started reloader process [10712] using StatReload
2023-04-19 03:25:13.691430: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 160563200 exceeds 10% of free system memory.
2023-04-19 03:25:14.177428: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 160563200 exceeds 10% of free system memory.
2023-04-19 03:25:14.230110: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 160563200 exceeds 10% of free system memory.
2023-04-19 03:25:15.825289: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 160563200 exceeds 10% of free system memory.
2023-04-19 03:25:15.865302: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 160563200 exceeds 10% of free system memory.
INFO:     Started server process [20704]
INFO:     Waiting for application startup.
INFO:     Application startup complete.

or you can run the code by running py main.py in the terminal and you will receive an output similar to this:

2023-04-19 02:58:57.877659: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 160563200 exceeds 10% of free system memory.
2023-04-19 02:58:58.345451: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 160563200 exceeds 10% of free system memory.
2023-04-19 02:58:58.397330: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 160563200 exceeds 10% of free system memory.
2023-04-19 02:58:59.431456: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 160563200 exceeds 10% of free system memory.
2023-04-19 02:58:59.467924: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 160563200 exceeds 10% of free system memory.
INFO:     Started server process [16784]
INFO:     Waiting for application startup.
INFO:     Application startup complete.
INFO:     Uvicorn running on http://127.0.0.1:8000 (Press CTRL+C to quit)

Then you can open the url in the browser and receive an output:

{
  "Hello": "World"
}

To try the API, simply follow these steps:

1. http://127.0.0.1:8000/docs, you will get to the documentation page
2. Find the POST method and click Try it out
3. Input the file which is an image and execute it
4. (Optional) you can use the example input in the input directory