This is the graphical user interface for the OCCRP document classifier.
A demo of this UI is hosted here: https://josemcorderoc.github.io/occrp-document-classifier-ui/. The demo only provides classification results of three example documents. The classification results are pre-computed as the machine learning model is not live.
In order to run the actual model, clone this repo via git clone https://github.com/josemcorderoc/occrp-document-classifier-ui.git. Open frontend/index.html in a browser. It should look like this:
For setting up the backend to this, see backend/README.md.
Render prediction JSON (JM)Clear button (JM)Thumbnails (in JS) (Frederik)progress circle thingadd table of results (datatables.net)remove the white horizontal bar at the very top- Validation of input (max size/num of pages/format)
- Write some things:
- How to understand the thresholds, why sometimes all numbers are zero
- CNN
- Which classes do we distinguish
- test mobile version (at least the information should look more or less look okay)
set up Google Analytics.- Add cookie banner and info on imprint page
- add imprint page
- add JPG support
- predicted other vs. predicted 0 have to be more clear in chart and bar chart
- columns in table and bar chart more clear
- when clicking on multiple Details, only one bar graph should be visible at one time
- design bar chart (probabilites in %, center bar chart)
- expand/collapse table has to be layouted (paddings) and is buggy at the moment, first time it doesn't collapse. When closed it should only read Expand, otherwise Collapse.
- result section should not be displayed in the beginning
probably add DSSG logo at the bottom- add attribution (sources) of images everywhere, below the images or in the Imprint. links are in the css file
- contact: maybe adding a contact formular? or give out our e-mails, but then might need some spam protection