README.md

---
language:
- en
base_model:
- openai/clip-vit-large-patch14
tags:
- memorability
- computer_vision
- perceptual_tasks
- CLIP
- LaMem
- THINGS
---
**PerceptCLIP-Memorability** is a model designed to predict **image memorability** (the likelihood of an image to be remembered). This is the official model from the paper:  
📄 **["Don't Judge Before You CLIP: A Unified Approach for Perceptual Tasks"](https://arxiv.org/abs/2503.13260)**.
We apply **LoRA adaptation** on the **CLIP visual encoder** and add an **MLP head** for memorability prediction. Our model achieves **state-of-the-art results**.

## Training Details

- *Dataset*: [LaMem](http://memorability.csail.mit.edu/download.html) (Large-Scale Image Memorability)
- *Architecture*: CLIP Vision Encoder (ViT-L/14) with *LoRA adaptation*
- *Loss Function*: Mean Squared Error (MSE) Loss for memorability prediction
- *Optimizer*: AdamW
- *Learning Rate*: 5e-05
- *Batch Size*: 32

## Installation & Requirements

You can set up the environment using environment.yml or manually install dependencies:
- python=3.9.15
- cudatoolkit=11.7
- torchvision=0.14.0
- transformers=4.45.2
- peft=0.14.0

## Usage

To use the model for inference:

```python
from torchvision import transforms
import torch
from PIL import Image
from huggingface_hub import hf_hub_download
import importlib.util

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

# Load the model class definition dynamically
class_path = hf_hub_download(repo_id="PerceptCLIP/PerceptCLIP_Memorability", filename="modeling.py")
spec = importlib.util.spec_from_file_location("modeling", class_path)
modeling = importlib.util.module_from_spec(spec)
spec.loader.exec_module(modeling)

# initialize a model
ModelClass = modeling.clip_lora_model 
model = ModelClass().to(device)

# Load pretrained model
model_path = hf_hub_download(repo_id="PerceptCLIP/PerceptCLIP_Memorability", filename="perceptCLIP_Memorability.pth")
model.load_state_dict(torch.load(model_path, map_location=device))
model.eval()
# Load an image
image = Image.open("image_path.jpg").convert("RGB")

# Preprocess and predict
def Mem_preprocess():
    transform = transforms.Compose([
        transforms.Resize(224),
        transforms.CenterCrop(size=(224, 224)),  
        transforms.ToTensor(),
        transforms.Normalize(mean=(0.48145466, 0.4578275, 0.40821073), 
                             std=(0.26862954, 0.26130258, 0.27577711))
    ])
    return transform

image = Mem_preprocess()(image).unsqueeze(0).to(device)

with torch.no_grad():
    mem_score = model(image).item()

print(f"Predicted Memorability Score: {mem_score:.4f}")
```

## Citation

If you use this model in your research, please cite:

```bibtex
@article{zalcher2025don,
  title={Don't Judge Before You CLIP: A Unified Approach for Perceptual Tasks},
  author={Zalcher, Amit and Wasserman, Navve and Beliy, Roman and Heinimann, Oliver and Irani, Michal},
  journal={arXiv preprint arXiv:2503.13260},
  year={2025}
}