Install Streamlit

pip install streamlit 

Downloading the unannotated image dataset

1. Download assigned dataset here. Unzip if neccessary.
2. Create a folder under any directory named results, then move the downloaded dataset folder(s) into the results directory.
3. Set the "root_path" in configs.json to be the directory containing the results/ folder containing

Run streamlit app

4. Clone this repo. Make sure your current directory is under annotation-ui/.

Your current directory should look like

- /synth-set-annotation-ui
  - app.py
  - configs.json
  - /annotation_res (generated after launch)
  - /results
    - /dataset1_name
      - /images
        - wg_0_0_dalle3.jpeg
        - ...
      - dataset1_name-metadata.json
    - ...

Then to launch annotation UI, run:

streamlit run app.py

Navigating the Streamlit UI

You could return to this page at any time by clicking the 'Return home' button or refreshing the page. Your annotation progress will be saved.

Now you can view and download your annotation results in the path:

./annotation_res/<selected dataset path>-res.json

Once you are finished annotating, please upload your entire annotation results folder to the following Google Drive folder and rename it to annotation_res-<your_name>.

./annotation_res/

I hope this makes things easier! ~emily