Awesome

Improving Zero-Shot Models with Label Distribution Priors

CLIPPR

Improving Zero-Shot Models with Label Distribution Priors
Joanthan Kahana, Niv Cohen, Yedid Hoshen
Official PyTorch Implementation

Abstract: Labeling large image datasets with attributes such as facial age or object type is tedious and sometimes infeasible. Supervised machine learning methods provide a highly accurate solution, but require manual labels which are often unavailable. Zero-shot models (e.g., CLIP) do not require manual labels but are not as accurate as supervised ones, particularly when the attribute is numeric. We propose a new approach, CLIPPR (CLIP with Priors), which adapts zero-shot models for regression and classification on unlabelled datasets. Our method does not use any annotated images. Instead, we assume a prior over the label distribution in the dataset. We then train an adapter network on top of CLIP under two competing objectives: i) minimal change of predictions from the original CLIP model ii) minimal distance between predicted and prior distribution of labels. Additionally, we present a novel approach for selecting prompts for Vision & Language models using a distributional prior. Our method is effective and presents a significant improvement over the original model. We demonstrate an improvement of $27%$ in mean absolute error on the UTK age regression task. We also present promising results for classification benchmarks, improving the classification accuracy on the ImageNet dataset by $1.83%$, without using any labels.

This repository is the official PyTorch implementation of Improving Zero-Shot Models with Label Distribution Priors

<a href="https://arxiv.org/abs/2212.00784" target="_blank"><img src="https://img.shields.io/badge/arXiv-2212.00784-b31b1b.svg"></a>

You can check out our project page here.

Usage

Requirements

Downloading the Datasets

You need to download the datasets first. download each to a separate directory under the same father directory.

NOTE: Please update the DATA_PATH parameter in dataset.py and scripts/prepare_stanford_cars.py to the father directory of the dataset.

For the ImageNet dataset please perform the pre-processing script found here.

For the Stanford Cars dataset please perform our pre-processing script: scripts/prepare_stanford_cars.py.

Training

We provide training & evaluation scripts for: CLIPPR, Zero-Shot CLIP (evaluation only) and a Supervised adapter on top of CLIP, for each one of the evaluated datasets.
The scripts can be found in the bash_scripts folder sorted by dataset.

Inside the bash_scripts\utk folder you can also find code for our ablation studies on that dataset.

For the imagenet dataset, please run first the warmup experiment bash_scripts\imagenet\inet_pretrain.sh, and only then the CLIPPR experiment.

NOTE: We also provide for running our experiments using sbatch array in the folder array_files.

Trained Checkpoints

Trained checkpoints can be found at this link.

Citation

If you find this useful, please cite our paper:

@article{kahana2022clippr,
  title={Improving Zero-Shot Models with Label Distribution Priors},
  author={Kahana, Jonathan and Cohen, Niv and Hoshen, Yedid},
  journal={arXiv preprint arXiv:2212.00784},
  year={2022}
}