This project hosts a Streamlit web application designed to classify objects in images using a pre-trained model. Users can upload images, and the application will display the images along with their predicted classes, and a visual distribution of detection results.
Features:
- Dark and light theme toggling for the application interface (preliminary version).
- Image classification using a pretrained Pytorch model.
- Visualization of the prediction results through an integrated Plotly graph:
Before running this application, make sure you have Docker installed on your system. If you do not have Docker, you can download it from the official Docker website.
To clone the repository and navigate into the project directory, run:
git clone https://github.com/PlanktoScope/streamlit-classification-app-with-fastapi-support.git
cd streamlit-classification-app-with-fastapi-support
To avoid errors later in running the docker container, please import your pretrained models (be sure to include the input image size in its name like sizexsize) and create a folder as follows:
models/<model_name>
By default, models must match Torchvision's efficientnet_v2_m architecture; to use models with a
different architecture, you must specify the architecture with a TORCHVISION_MODEL_TYPE
environment variable passed into the app.
Build the Docker image using the following command (it takes a considerable time):
docker build -t <image_name>:<tag> .
Run the Docker container with:
docker run -p 8501:8501 <image_name>:<tag>
The Streamlit app will be served on port 8501.
Once the application is running, click on the link displayed in your terminal or open your web browser and navigate to:
http://localhost:8501
From there, you can use the web interface to:
- Upload images for classification.
- View model predictions and the distribution of detected objects.
This version of the Streamlit web app contains the code for a FastAPI server hosting the model selected for the classification of mussel and oyster larvae. This model was trained on hatchery larvae images.
Starting the FastAPI Server:
- Navigate to the fastapi folder within the fastapi model serving version:
cd fastapi_model_serving/fastapi
- Run the FastAPI server:
uvicorn server:api_server --host 0.0.0.0 --port 8000
You can use Forklift to easily deploy the Docker container provided by this repository.
You will need to have the Docker Engine installed on your computer. Installation instructions are available here.
Then, you will need to set up the forklift tool on
your computer. Setup instructions are available
here. Note
that currently forklift is only tested for Linux computers, and that Forklift is still an
experimental prototype.
You can clone, stage, and apply the latest commit of this Forklift pallet to your computer, by
using the forklift tool:
sudo -E forklift pallet switch --apply github.com/PlanktoScope/streamlit-classification-app@main
Warning: this will replace all Docker containers on your Docker host with the package deployments specified by this pallet and delete any Docker containers not specified by this pallet's package deployments.
If your user is in the docker group
(so that you don't need to use sudo when running docker commands), then you don't need to use
sudo with forklift:
forklift pallet switch --apply github.com/PlanktoScope/streamlit-classification-app@main
When this pallet is updated on GitHub and you want to apply the latest changes on your computer, you
can use the same command as above (either
forklift pallet switch --apply github.com/PlanktoScope/streamlit-classification-app@main or that command with
sudo -E) to clone, stage, and apply the updated version of the pallet.
Because the forklift tool uses Docker Compose to manage the
Docker containers specified by this pallet, the containers will not be running after you restart
your computer (this is true each time you restart your computer); you will need to run a command to
start the containers again:
sudo -E forklift stage apply
Or if your user is in the docker group:
forklift stage apply
After you have applied the pallet so that the streamlit demo app's container is running, you can access the streamlit demo app from your web browser at http://localhost/ps/streamlit-demo.
Before you can use the demo app, you will need to download a classification model weights file
(e.g. https://github.com/PlanktoScope/streamlit-classification-app/releases/download/models%2Fdemo-1/effv2s_no_norm_DA+sh_20patience_256x256_50ep_loss.pth)
into ~/.local/share/planktoscope/models; by default the model weights file must be for the
efficientnet_v2_s model architecture, but you can use the efficientnet_v2_m model architecture
instead by disabling the torchvision-model-efficientnet-v2-s feature flag of the pallet's
apps/ps/streamlit-demo package deployment.
Then you can upload input images (e.g. https://github.com/PlanktoScope/streamlit-classification-app/releases/download/models%2Fdemo-1/example-input-tots-ps-acq-20-02_49_37_288982.jpg) to the demo app.
This project is licensed under the Apache-2.0.
Copyright Wassim Chakroun and PlanktoScope project contributors.