|
|
|
Directory Overview: This directory contains all the evaluation modules and results: |
|
|
|
################################################################################################################################ |
|
|
|
## 1. evaluation.py |
|
The `evaluation.py` module provides a comprehensive framework for evaluating Knowledge-Based Visual Question |
|
Answering (KB-VQA) models. This module includes methods for syntactic and semantic evaluation, leveraging both |
|
exact match and VQA scores. The evaluation results can be saved to an Excel file for further analysis. |
|
|
|
Notes |
|
Ensure that the necessary configuration settings are correctly set in "my_model/config/evaluation_config" file. |
|
The module requires the following dependencies: pandas, fuzzywuzzy, nltk, ast, typing, streamlit, and openai. |
|
|
|
################################################################################################################################ |
|
|
|
## 2. demo.py |
|
The `demo.py` module includes the `ResultDemonstrator` class, which provides a comprehensive framework for visualizing |
|
and demonstrating the results of Knowledge-Based Visual Question Answering (KB-VQA) models. |
|
The class utilizes Streamlit for displaying data and Altair for creating interactive visualizations. |
|
this class is the backbone for the results demo tool of the application. |
|
|
|
Notes |
|
Ensure that the necessary configuration settings are correctly set in "my_model/config/evaluation_config". |
|
The module requires the following dependencies: os, altair, streamlit, PIL, pandas, random. |
|
|
|
################################################################################################################################ |
|
|
|
## 3. evaluation_results.xlsx |
|
This file contains all the evaluation results based on the evaluation data (~ 1,000 samples), the results are in the sheet names "Main Data". |
|
running the evaluation module will read the main columns (Model Answers for all model's configurations) from this sheet and regenerate and save the evaluation rsults. |
|
This Main Data sheet is the result of running the evaluation module (with the same file name chosen for save). |
|
|
|
############################################################################################################################### |
|
|
|
## 4. Demo_Images |
|
This folder contains a pool of images used for the demo app, to visualize and demonstrate the images together with the model answers. |