Edit model card

Chart Recognizer

chart-recognizer is a finetuned model for classifying images. It uses efficientnet as its base model, making it a fast and small model. This model is trained on my own dataset of financial charts posted on Twitter, which can be found here StephanAkkerman/fintwit-charts.

Intended Uses

chart-recognizer is intended for classifying images, mainly images posted on social media.

Dataset

chart-recognizer has been trained on my own dataset. So far I have not been able to find another image dataset about financial charts.

Example Images

The following images are not part of the training set and can be used for testing purposes.

Chart

image/png

Non-Chart

This can be any image that does not represent a (financial) chart. image/png

More Information

For a comprehensive overview, including the training setup and analysis of the model, visit the chart-recognizer GitHub repository.

Usage

Using HuggingFace's transformers library the model can be converted into a pipeline for image classification.

import timm
import torch
from PIL import Image
from timm.data import resolve_data_config, create_transform

# Load and set model to eval mode
model = timm.create_model("hf_hub:StephanAkkerman/chart-recognizer", pretrained=True)
model.eval()

# Create transform and get labels
transform = create_transform(**resolve_data_config(model.pretrained_cfg, model=model))
labels = model.pretrained_cfg["label_names"]

# Load and preprocess image
image = Image.open("img/examples/tweet_example.png").convert("RGB")
x = transform(image).unsqueeze(0)

# Get model output and apply softmax
probabilities = torch.nn.functional.softmax(model(x)[0], dim=0)

# Map probabilities to labels
output = {label: prob.item() for label, prob in zip(labels, probabilities)}

# Print the predicted probabilities
print(output)

Citing & Authors

If you use chart-recognizer in your research, please cite me as follows:

@misc{chart-recognizer,
  author = {Stephan Akkerman},
  title = {chart-recognizer: A Specialized Image Model for Financial Charts},
  year = {2024},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/StephanAkkerman/chart-recognizer}}
}

License

This project is licensed under the MIT License. See the LICENSE file for details.

Downloads last month
178

Finetuned from

Datasets used to train StephanAkkerman/chart-recognizer

Evaluation results