michelecafagna26

faster-rcnn-bottom-up-py

Extract features and bounding boxes using the original Bottom-up Attention Faster-RCNN in a few lines of Python code

cider

Pythonic wrappers for Cider/CiderD evaluation metrics. Provides CIDEr as well as CIDEr-D (CIDEr Defended) which is more robust to gaming effects. We also add the possibility to replace the original PTBTokenizer with the Spacy tekenizer (No java dependincy but slower)

VinVL

Original VinVL (and Oscar) repo with API designed for an easy inference

vinvl-visualbackbone

Original VinVL visual backbone with simplified APIs to easily extract features, boxes, object detections, in a few lines of Python code.

CHANGE-IT

[EVALITA2020] CHANGE-IT Task

CLIPCap

ClipCap implementation with API for an easy inference + baselines for the HL Dataset

HL-dataset

[INLG2023] The High-Level (HL) dataset is a Vision and Language (V&L) resource aligning object-centric descriptions from COCO with high-level descriptions crowdsourced along 3 axes: scene, action, rationale.

vl-shap

[Frontiers in AI Journal] Implementation of the paper "Interpreting Vision and Language Generative Models with Semantic Visual Priors"

bibidy

A simple command line tool for basic manipulations of bib files.

compress-fasttext

Tools for shrinking fastText models (in gensim format)

ELFVisionModule

ELF Vision Module demo

michelecafagna26

Github profile page

michelecafagna26.github.io

Pascal_Sentence_Dataset-downloader

Utility to downlaod the Pascal Sentence Dataset

text_completion_api

A simple text completion API using GPT-J-6B

vl-ablation

Targeted semantic multimodal input ablation. Official implementation of the ablation method introduced in the paper: "What Vision-Language Models 'See' when they See Scenes"

vl-shap-demo

Gradio demo showcasing VL-SHAP. You can generate visually informed explanations of textual outputs for VL models