Skip to content

updated visualBERT modelcard #40057

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
stevhliu merged 2 commits into from
Aug 13, 2025
Merged

updated visualBERT modelcard #40057

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
1ad044c
updated visualBERT modelcard
Anil-Red Aug 9, 2025
149c13c
fix: Review for VisualBERT card
Anil-Red Aug 13, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
175 changes: 102 additions & 73 deletions docs/source/en/model_doc/visual_bert.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -14,87 +14,116 @@ rendered properly in your Markdown viewer.

-->

# VisualBERT

<div class="flex flex-wrap space-x-1">
<img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
<div style="float: right;">
Comment thread
Anil-Red marked this conversation as resolved.
<div class="flex flex-wrap space-x-1">
<img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
</div>
</div>

## Overview

The VisualBERT model was proposed in [VisualBERT: A Simple and Performant Baseline for Vision and Language](https://huggingface.co/papers/1908.03557) by Liunian Harold Li, Mark Yatskar, Da Yin, Cho-Jui Hsieh, Kai-Wei Chang.
VisualBERT is a neural network trained on a variety of (image, text) pairs.

The abstract from the paper is the following:

*We propose VisualBERT, a simple and flexible framework for modeling a broad range of vision-and-language tasks.
VisualBERT consists of a stack of Transformer layers that implicitly align elements of an input text and regions in an
associated input image with self-attention. We further propose two visually-grounded language model objectives for
pre-training VisualBERT on image caption data. Experiments on four vision-and-language tasks including VQA, VCR, NLVR2,
and Flickr30K show that VisualBERT outperforms or rivals with state-of-the-art models while being significantly
simpler. Further analysis demonstrates that VisualBERT can ground elements of language to image regions without any
explicit supervision and is even sensitive to syntactic relationships, tracking, for example, associations between
verbs and image regions corresponding to their arguments.*

This model was contributed by [gchhablani](https://huggingface.co/gchhablani). The original code can be found [here](https://github.com/uclanlp/visualbert).

## Usage tips

1. Most of the checkpoints provided work with the [`VisualBertForPreTraining`] configuration. Other
checkpoints provided are the fine-tuned checkpoints for down-stream tasks - VQA ('visualbert-vqa'), VCR
('visualbert-vcr'), NLVR2 ('visualbert-nlvr2'). Hence, if you are not working on these downstream tasks, it is
recommended that you use the pretrained checkpoints.

2. For the VCR task, the authors use a fine-tuned detector for generating visual embeddings, for all the checkpoints.
We do not provide the detector and its weights as a part of the package, but it will be available in the research
projects, and the states can be loaded directly into the detector provided.

VisualBERT is a multi-modal vision and language model. It can be used for visual question answering, multiple choice,
visual reasoning and region-to-phrase correspondence tasks. VisualBERT uses a BERT-like transformer to prepare
embeddings for image-text pairs. Both the text and visual features are then projected to a latent space with identical
dimension.

To feed images to the model, each image is passed through a pre-trained object detector and the regions and the
bounding boxes are extracted. The authors use the features generated after passing these regions through a pre-trained
CNN like ResNet as visual embeddings. They also add absolute position embeddings, and feed the resulting sequence of
vectors to a standard BERT model. The text input is concatenated in the front of the visual embeddings in the embedding
layer, and is expected to be bound by [CLS] and a [SEP] tokens, as in BERT. The segment IDs must also be set
appropriately for the textual and visual parts.

The [`BertTokenizer`] is used to encode the text. A custom detector/image processor must be used
to get the visual embeddings. The following example notebooks show how to use VisualBERT with Detectron-like models:

- [VisualBERT VQA demo notebook](https://github.com/huggingface/transformers-research-projects/tree/main/visual_bert) : This notebook
contains an example on VisualBERT VQA.
# VisualBERT

- [Generate Embeddings for VisualBERT (Colab Notebook)](https://colab.research.google.com/drive/1bLGxKdldwqnMVA5x4neY7-l_8fKGWQYI?usp=sharing) : This notebook contains
an example on how to generate visual embeddings.
[VisualBERT](https://huggingface.co/papers/1908.03557) is a vision-and-language model. It uses an approach called "early fusion", where inputs are fed together into a single Transformer stack initialized from [BERT](./bert). Self-attention implicitly aligns words with their corresponding image objects. It processes text with visual features from object-detector regions instead of raw pixels.

You can find all the original VisualBERT checkpoints under the [UCLA NLP](https://huggingface.co/uclanlp/models?search=visualbert) organization.


> [!TIP]
Comment thread
stevhliu marked this conversation as resolved.
> This model was contributed by [gchhablani](https://huggingface.co/gchhablani).
> Click on the VisualBERT models in the right sidebar for more examples of how to apply VisualBERT to different image and language tasks.

The example below demonstrates how to answer a question based on an image with the [`AutoModel`] class.

<hfoptions id="usage">
<hfoption id="AutoModel">
Comment thread
stevhliu marked this conversation as resolved.

```py
import torch
import torchvision
from PIL import Image
import numpy as np
from transformers import AutoTokenizer, VisualBertForQuestionAnswering
import requests
from io import BytesIO

def get_visual_embeddings_simple(image, device=None):

model = torchvision.models.resnet50(pretrained=True)
model = torch.nn.Sequential(*list(model.children())[:-1])
model.to(device)
model.eval()

transform = torchvision.transforms.Compose([
torchvision.transforms.Resize(256),
torchvision.transforms.CenterCrop(224),
torchvision.transforms.ToTensor(),
torchvision.transforms.Normalize(
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]
)
])

if isinstance(image, str):
image = Image.open(image).convert('RGB')
elif isinstance(image, Image.Image):
image = image.convert('RGB')
else:
raise ValueError("Image must be a PIL Image or path to image file")

image_tensor = transform(image).unsqueeze(0).to(device)

with torch.no_grad():
features = model(image_tensor)

batch_size = features.shape[0]
feature_dim = features.shape[1]
visual_seq_length = 10

visual_embeds = features.squeeze(-1).squeeze(-1).unsqueeze(1).expand(batch_size, visual_seq_length, feature_dim)

return visual_embeds

tokenizer = AutoTokenizer.from_pretrained("google-bert/bert-base-uncased")
model = VisualBertForQuestionAnswering.from_pretrained("uclanlp/visualbert-vqa-coco-pre")

response = requests.get("https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg")
image = Image.open(BytesIO(response.content))

visual_embeds = get_visual_embeddings_simple(image)

inputs = tokenizer("What is shown in this image?", return_tensors="pt")

visual_token_type_ids = torch.ones(visual_embeds.shape[:-1], dtype=torch.long)
visual_attention_mask = torch.ones(visual_embeds.shape[:-1], dtype=torch.float)

inputs.update({
"visual_embeds": visual_embeds,
"visual_token_type_ids": visual_token_type_ids,
"visual_attention_mask": visual_attention_mask,
})

with torch.no_grad():
outputs = model(**inputs)
logits = outputs.logits
predicted_answer_idx = logits.argmax(-1).item()

print(f"Predicted answer: {predicted_answer_idx}")
```

The following example shows how to get the last hidden state using [`VisualBertModel`]:
</hfoption>
</hfoptions>

```python
>>> import torch
>>> from transformers import BertTokenizer, VisualBertModel
## Notes

>>> model = VisualBertModel.from_pretrained("uclanlp/visualbert-vqa-coco-pre")
>>> tokenizer = BertTokenizer.from_pretrained("google-bert/bert-base-uncased")
- Use a fine-tuned checkpoint for downstream tasks, like `visualbert-vqa` for visual question answering. Otherwise, use one of the pretrained checkpoints.
- The fine-tuned detector and weights aren't provided (available in the research projects), but the states can be directly loaded into the detector.
- The text input is concatenated in front of the visual embeddings in the embedding layer and is expected to be bound by `[CLS]` and [`SEP`] tokens.
- The segment ids must be set appropriately for the text and visual parts.
- Use [`BertTokenizer`] to encode the text and implement a custom detector/image processor to get the visual embeddings.

>>> inputs = tokenizer("What is the man eating?", return_tensors="pt")
>>> # this is a custom function that returns the visual embeddings given the image path
>>> visual_embeds = get_visual_embeddings(image_path)
## Resources

>>> visual_token_type_ids = torch.ones(visual_embeds.shape[:-1], dtype=torch.long)
>>> visual_attention_mask = torch.ones(visual_embeds.shape[:-1], dtype=torch.float)
>>> inputs.update(
... {
... "visual_embeds": visual_embeds,
... "visual_token_type_ids": visual_token_type_ids,
... "visual_attention_mask": visual_attention_mask,
... }
... )
>>> outputs = model(**inputs)
>>> last_hidden_state = outputs.last_hidden_state
```
- Refer to this [notebook](https://github.com/huggingface/transformers-research-projects/tree/main/visual_bert) for an example of using VisualBERT for visual question answering.
- Refer to this [notebook](https://colab.research.google.com/drive/1bLGxKdldwqnMVA5x4neY7-l_8fKGWQYI?usp=sharing) for an example of how to generate visual embeddings.

## VisualBertConfig

Expand Down