🛰️ Official repository of paper "RemoteCLIP: A Vision Language Foundation Model for Remote Sensing" (IEEE TGRS)

An extremely simple method for validation-free efficient adaptation of CLIP-like VLMs that is robust to the learning rate.

[ICLR2025] Detecting Backdoor Samples in Contrastive Language Image Pretraining

Use CLIP to create matching texts + embeddings for given images; useful for XAI, adversarial training

Simple data and training pipeline for class-incremental method 😄

CLIP-interrogator InvokeAI node

Theory, Experiments, and Dataset for our newly proposed Deep Learning method for LLM-driven Cycle Consistency and Semantics Aware Self-Supervised Framework for Unpaired LDR ↔ HDR Image Translation

Visual and Vision-Language Representation Pre-Training with Contrastive Learning

This application fine-tunes the CLIP model on the Flickr8k dataset to align image and text embeddings for image-caption matching. It includes a clean and reproducible pipeline for data preparation, model training, evaluation, and inference. The goal is to enhance multimodal understanding and retrieval performance using a custom captioning dataset.

Attempts to improve CLIP via different optimizers and loss functions

University project based on implementing a Test Time Adaptation (TTA) solution for image classifiers. University of Trento (Italy)