Kye
1 year ago
parent
d33588becc
commit
00448fa20e
@ -1,183 +0,0 @@
|
||||
"""
|
||||
|
||||
|
||||
BiomedCLIP-PubMedBERT_256-vit_base_patch16_224
|
||||
https://huggingface.co/microsoft/BiomedCLIP-PubMedBERT_256-vit_base_patch16_224
|
||||
BiomedCLIP is a biomedical vision-language foundation model that is pretrained on PMC-15M,
|
||||
a dataset of 15 million figure-caption pairs extracted from biomedical research articles in PubMed Central, using contrastive learning. It uses PubMedBERT as the text encoder and Vision Transformer as the image encoder, with domain-specific adaptations. It can perform various vision-language processing (VLP) tasks such as cross-modal retrieval, image classification, and visual question answering. BiomedCLIP establishes new state of the art in a wide range of standard datasets, and substantially outperforms prior VLP approaches:
|
||||
|
||||
|
||||
|
||||
Citation
|
||||
@misc{https://doi.org/10.48550/arXiv.2303.00915,
|
||||
doi = {10.48550/ARXIV.2303.00915},
|
||||
url = {https://arxiv.org/abs/2303.00915},
|
||||
author = {Zhang, Sheng and Xu, Yanbo and Usuyama, Naoto and Bagga, Jaspreet and Tinn, Robert and Preston, Sam and Rao, Rajesh and Wei, Mu and Valluri, Naveen and Wong, Cliff and Lungren, Matthew and Naumann, Tristan and Poon, Hoifung},
|
||||
title = {Large-Scale Domain-Specific Pretraining for Biomedical Vision-Language Processing},
|
||||
publisher = {arXiv},
|
||||
year = {2023},
|
||||
}
|
||||
|
||||
Model Use
|
||||
How to use
|
||||
Please refer to this example notebook.
|
||||
|
||||
Intended Use
|
||||
This model is intended to be used solely for (I) future research on visual-language processing and (II) reproducibility of the experimental results reported in the reference paper.
|
||||
|
||||
Primary Intended Use
|
||||
The primary intended use is to support AI researchers building on top of this work. BiomedCLIP and its associated models should be helpful for exploring various biomedical VLP research questions, especially in the radiology domain.
|
||||
|
||||
Out-of-Scope Use
|
||||
Any deployed use case of the model --- commercial or otherwise --- is currently out of scope. Although we evaluated the models using a broad set of publicly-available research benchmarks, the models and evaluations are not intended for deployed use cases. Please refer to the associated paper for more details.
|
||||
|
||||
Data
|
||||
This model builds upon PMC-15M dataset, which is a large-scale parallel image-text dataset for biomedical vision-language processing. It contains 15 million figure-caption pairs extracted from biomedical research articles in PubMed Central. It covers a diverse range of biomedical image types, such as microscopy, radiography, histology, and more.
|
||||
|
||||
Limitations
|
||||
This model was developed using English corpora, and thus can be considered English-only.
|
||||
|
||||
Further information
|
||||
Please refer to the corresponding paper, "Large-Scale Domain-Specific Pretraining for Biomedical Vision-Language Processing" for additional details on the model training and evaluation.
|
||||
"""
|
||||
|
||||
import open_clip
|
||||
import torch
|
||||
from PIL import Image
|
||||
import matplotlib.pyplot as plt
|
||||
|
||||
|
||||
class BioClip:
|
||||
"""
|
||||
BioClip
|
||||
|
||||
Args:
|
||||
model_path (str): path to the model
|
||||
|
||||
Attributes:
|
||||
model_path (str): path to the model
|
||||
model (torch.nn.Module): the model
|
||||
preprocess_train (torchvision.transforms.Compose): the preprocessing pipeline for training
|
||||
preprocess_val (torchvision.transforms.Compose): the preprocessing pipeline for validation
|
||||
tokenizer (open_clip.Tokenizer): the tokenizer
|
||||
device (torch.device): the device to run the model on
|
||||
|
||||
Methods:
|
||||
__call__(self, img_path: str, labels: list, template: str = 'this is a photo of ', context_length: int = 256):
|
||||
returns a dictionary of labels and their probabilities
|
||||
plot_image_with_metadata(img_path: str, metadata: dict): plots the image with the metadata
|
||||
|
||||
Usage:
|
||||
clip = BioClip('hf-hub:microsoft/BiomedCLIP-PubMedBERT_256-vit_base_patch16_224')
|
||||
|
||||
labels = [
|
||||
'adenocarcinoma histopathology',
|
||||
'brain MRI',
|
||||
'covid line chart',
|
||||
'squamous cell carcinoma histopathology',
|
||||
'immunohistochemistry histopathology',
|
||||
'bone X-ray',
|
||||
'chest X-ray',
|
||||
'pie chart',
|
||||
'hematoxylin and eosin histopathology'
|
||||
]
|
||||
|
||||
result = clip("your_image_path.jpg", labels)
|
||||
metadata = {'filename': "your_image_path.jpg".split('/')[-1], 'top_probs': result}
|
||||
clip.plot_image_with_metadata("your_image_path.jpg", metadata)
|
||||
|
||||
|
||||
"""
|
||||
|
||||
def __init__(self, model_path: str):
|
||||
self.model_path = model_path
|
||||
(
|
||||
self.model,
|
||||
self.preprocess_train,
|
||||
self.preprocess_val,
|
||||
) = open_clip.create_model_and_transforms(model_path)
|
||||
self.tokenizer = open_clip.get_tokenizer(model_path)
|
||||
self.device = (
|
||||
torch.device("cuda")
|
||||
if torch.cuda.is_available()
|
||||
else torch.device("cpu")
|
||||
)
|
||||
self.model.to(self.device)
|
||||
self.model.eval()
|
||||
|
||||
def __call__(
|
||||
self,
|
||||
img_path: str,
|
||||
labels: list,
|
||||
template: str = "this is a photo of ",
|
||||
context_length: int = 256,
|
||||
):
|
||||
image = torch.stack(
|
||||
[self.preprocess_val(Image.open(img_path))]
|
||||
).to(self.device)
|
||||
texts = self.tokenizer(
|
||||
[template + l for l in labels],
|
||||
context_length=context_length,
|
||||
).to(self.device)
|
||||
|
||||
with torch.no_grad():
|
||||
image_features, text_features, logit_scale = self.model(
|
||||
image, texts
|
||||
)
|
||||
logits = (
|
||||
(logit_scale * image_features @ text_features.t())
|
||||
.detach()
|
||||
.softmax(dim=-1)
|
||||
)
|
||||
sorted_indices = torch.argsort(
|
||||
logits, dim=-1, descending=True
|
||||
)
|
||||
logits = logits.cpu().numpy()
|
||||
sorted_indices = sorted_indices.cpu().numpy()
|
||||
|
||||
results = {}
|
||||
for idx in sorted_indices[0]:
|
||||
label = labels[idx]
|
||||
prob = logits[0][idx]
|
||||
results[label] = prob
|
||||
return results
|
||||
|
||||
@staticmethod
|
||||
def plot_image_with_metadata(img_path: str, metadata: dict):
|
||||
img = Image.open(img_path)
|
||||
fig, ax = plt.subplots(figsize=(5, 5))
|
||||
ax.imshow(img)
|
||||
ax.axis("off")
|
||||
title = (
|
||||
metadata["filename"]
|
||||
+ "\n"
|
||||
+ "\n".join(
|
||||
[
|
||||
f"{k}: {v*100:.1f}"
|
||||
for k, v in metadata["top_probs"].items()
|
||||
]
|
||||
)
|
||||
)
|
||||
ax.set_title(title, fontsize=14)
|
||||
plt.tight_layout()
|
||||
plt.show()
|
||||
|
||||
|
||||
# Usage
|
||||
# clip = BioClip('hf-hub:microsoft/BiomedCLIP-PubMedBERT_256-vit_base_patch16_224')
|
||||
|
||||
# labels = [
|
||||
# 'adenocarcinoma histopathology',
|
||||
# 'brain MRI',
|
||||
# 'covid line chart',
|
||||
# 'squamous cell carcinoma histopathology',
|
||||
# 'immunohistochemistry histopathology',
|
||||
# 'bone X-ray',
|
||||
# 'chest X-ray',
|
||||
# 'pie chart',
|
||||
# 'hematoxylin and eosin histopathology'
|
||||
# ]
|
||||
|
||||
# result = clip("your_image_path.jpg", labels)
|
||||
# metadata = {'filename': "your_image_path.jpg".split('/')[-1], 'top_probs': result}
|
||||
# clip.plot_image_with_metadata("your_image_path.jpg", metadata)
|
||||
@ -1,169 +0,0 @@
|
||||
# Import necessary modules and define fixtures if needed
|
||||
import pytest
|
||||
import torch
|
||||
from PIL import Image
|
||||
from swarms.models.bioclip import BioClip
|
||||
|
||||
|
||||
# Define fixtures if needed
|
||||
@pytest.fixture
|
||||
def sample_image_path():
|
||||
return "path_to_sample_image.jpg"
|
||||
|
||||
|
||||
@pytest.fixture
|
||||
def clip_instance():
|
||||
return BioClip(
|
||||
"microsoft/BiomedCLIP-PubMedBERT_256-vit_base_patch16_224"
|
||||
)
|
||||
|
||||
|
||||
# Basic tests for the BioClip class
|
||||
def test_clip_initialization(clip_instance):
|
||||
assert isinstance(clip_instance.model, torch.nn.Module)
|
||||
assert hasattr(clip_instance, "model_path")
|
||||
assert hasattr(clip_instance, "preprocess_train")
|
||||
assert hasattr(clip_instance, "preprocess_val")
|
||||
assert hasattr(clip_instance, "tokenizer")
|
||||
assert hasattr(clip_instance, "device")
|
||||
|
||||
|
||||
def test_clip_call_method(clip_instance, sample_image_path):
|
||||
labels = [
|
||||
"adenocarcinoma histopathology",
|
||||
"brain MRI",
|
||||
"covid line chart",
|
||||
"squamous cell carcinoma histopathology",
|
||||
"immunohistochemistry histopathology",
|
||||
"bone X-ray",
|
||||
"chest X-ray",
|
||||
"pie chart",
|
||||
"hematoxylin and eosin histopathology",
|
||||
]
|
||||
result = clip_instance(sample_image_path, labels)
|
||||
assert isinstance(result, dict)
|
||||
assert len(result) == len(labels)
|
||||
|
||||
|
||||
def test_clip_plot_image_with_metadata(
|
||||
clip_instance, sample_image_path
|
||||
):
|
||||
metadata = {
|
||||
"filename": "sample_image.jpg",
|
||||
"top_probs": {"label1": 0.75, "label2": 0.65},
|
||||
}
|
||||
clip_instance.plot_image_with_metadata(
|
||||
sample_image_path, metadata
|
||||
)
|
||||
|
||||
|
||||
# More test cases can be added to cover additional functionality and edge cases
|
||||
|
||||
|
||||
# Parameterized tests for different image and label combinations
|
||||
@pytest.mark.parametrize(
|
||||
"image_path, labels",
|
||||
[
|
||||
("image1.jpg", ["label1", "label2"]),
|
||||
("image2.jpg", ["label3", "label4"]),
|
||||
# Add more image and label combinations
|
||||
],
|
||||
)
|
||||
def test_clip_parameterized_calls(clip_instance, image_path, labels):
|
||||
result = clip_instance(image_path, labels)
|
||||
assert isinstance(result, dict)
|
||||
assert len(result) == len(labels)
|
||||
|
||||
|
||||
# Test image preprocessing
|
||||
def test_clip_image_preprocessing(clip_instance, sample_image_path):
|
||||
image = Image.open(sample_image_path)
|
||||
processed_image = clip_instance.preprocess_val(image)
|
||||
assert isinstance(processed_image, torch.Tensor)
|
||||
|
||||
|
||||
# Test label tokenization
|
||||
def test_clip_label_tokenization(clip_instance):
|
||||
labels = ["label1", "label2"]
|
||||
tokenized_labels = clip_instance.tokenizer(labels)
|
||||
assert isinstance(tokenized_labels, torch.Tensor)
|
||||
assert tokenized_labels.shape[0] == len(labels)
|
||||
|
||||
|
||||
# More tests can be added to cover other methods and edge cases
|
||||
|
||||
|
||||
# End-to-end tests with actual images and labels
|
||||
def test_clip_end_to_end(clip_instance, sample_image_path):
|
||||
labels = [
|
||||
"adenocarcinoma histopathology",
|
||||
"brain MRI",
|
||||
"covid line chart",
|
||||
"squamous cell carcinoma histopathology",
|
||||
"immunohistochemistry histopathology",
|
||||
"bone X-ray",
|
||||
"chest X-ray",
|
||||
"pie chart",
|
||||
"hematoxylin and eosin histopathology",
|
||||
]
|
||||
result = clip_instance(sample_image_path, labels)
|
||||
assert isinstance(result, dict)
|
||||
assert len(result) == len(labels)
|
||||
|
||||
|
||||
# Test label tokenization with long labels
|
||||
def test_clip_long_labels(clip_instance):
|
||||
labels = ["label" + str(i) for i in range(100)]
|
||||
tokenized_labels = clip_instance.tokenizer(labels)
|
||||
assert isinstance(tokenized_labels, torch.Tensor)
|
||||
assert tokenized_labels.shape[0] == len(labels)
|
||||
|
||||
|
||||
# Test handling of multiple image files
|
||||
def test_clip_multiple_images(clip_instance, sample_image_path):
|
||||
labels = ["label1", "label2"]
|
||||
image_paths = [sample_image_path, "image2.jpg"]
|
||||
results = clip_instance(image_paths, labels)
|
||||
assert isinstance(results, list)
|
||||
assert len(results) == len(image_paths)
|
||||
for result in results:
|
||||
assert isinstance(result, dict)
|
||||
assert len(result) == len(labels)
|
||||
|
||||
|
||||
# Test model inference performance
|
||||
def test_clip_inference_performance(
|
||||
clip_instance, sample_image_path, benchmark
|
||||
):
|
||||
labels = [
|
||||
"adenocarcinoma histopathology",
|
||||
"brain MRI",
|
||||
"covid line chart",
|
||||
"squamous cell carcinoma histopathology",
|
||||
"immunohistochemistry histopathology",
|
||||
"bone X-ray",
|
||||
"chest X-ray",
|
||||
"pie chart",
|
||||
"hematoxylin and eosin histopathology",
|
||||
]
|
||||
result = benchmark(clip_instance, sample_image_path, labels)
|
||||
assert isinstance(result, dict)
|
||||
assert len(result) == len(labels)
|
||||
|
||||
|
||||
# Test different preprocessing pipelines
|
||||
def test_clip_preprocessing_pipelines(
|
||||
clip_instance, sample_image_path
|
||||
):
|
||||
image = Image.open(sample_image_path)
|
||||
|
||||
# Test preprocessing for training
|
||||
processed_image_train = clip_instance.preprocess_train(image)
|
||||
assert isinstance(processed_image_train, torch.Tensor)
|
||||
|
||||
# Test preprocessing for validation
|
||||
processed_image_val = clip_instance.preprocess_val(image)
|
||||
assert isinstance(processed_image_val, torch.Tensor)
|
||||
|
||||
|
||||
# ...
|
||||
Loading…
Reference in new issue