---
license: apache-2.0
tags:
- image-captioning
languages:
- en
pipeline_tag: image-to-text
datasets:
- michelecafagna26/hl
language:
- en
metrics:
- sacrebleu
- rouge
library_name: transformers
---
## BLIP-base fine-tuned for Image Capioning on High-Level descriptions of Actons

[BLIP](https://arxiv.org/abs/2201.12086) base trained on the [HL dataset](https://huggingface.co/datasets/michelecafagna26/hl) for **high-level descriptions of actions**

## Model fine-tuning 🏋️‍

Trained for a maximum of 6 epochs using a
lr:  5e−5,
Adam optimizer,
half-precision (fp16)

## Test set metrics 🧾

    | Cider  | SacreBLEU  | Rouge-L|
    |--------|------------|--------|
    | 123.07 |    17.16   |  32.16 |

## Model in Action 🚀

```python
import requests
from PIL import Image
from transformers import BlipProcessor, BlipForConditionalGeneration

processor = BlipProcessor.from_pretrained("Salesforce/blip-image-captioning-base")
model = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-base").to("cuda")

img_url = 'https://datasets-server.huggingface.co/assets/michelecafagna26/hl/--/default/train/0/image/image.jpg' 
raw_image = Image.open(requests.get(img_url, stream=True).raw).convert('RGB')


inputs = processor(raw_image, return_tensors="pt").to("cuda")
pixel_values = inputs.pixel_values

generated_ids = model.generate(pixel_values=pixel_values, max_length=50,
            do_sample=True,
            top_k=120,
            top_p=0.9,
            early_stopping=True,
            num_return_sequences=1)

processor.batch_decode(generated_ids, skip_special_tokens=True)

>>> she's holding a parasol
```

## BibTex and citation info

```BibTeX
```