Image Emotion Detection

Emotion Detection via a Deep Learning Framework

[Abstract]

[Requirement]

• Python >= 3.x
• Scikit-learn https://scikit-learn.org/stable/
• Numpy https://numpy.org/
• Keras https://keras.io/
• CV2 https://pypi.org/project/opencv-python/

[Data]

• data/pretrained_data: only a subset of sampled images. For the full set, please email: kpzhang@umd.edu or refer to: https://dl.acm.org/doi/10.5555/3015812.3015857
• data/kickstarter_data: images of all crowdfunding projects for this study

[Trained Model]

The trained model can be downloaded here

For other models from the baselines, please contact us.

[Run]

• get_objects_google.py: using google vision API to obtain objects embedded in an image

• go to ANP folder for codes to obtain ANPs for each image

• FPN_train.py: Feature Pyramid Network

• train.py: training procedure of our model

• predict.py: predicting image emotions

• xgboost.py: one of baselines using XGBoost

• To obtain the ANPs, please refer to the details here: https://www.ee.columbia.edu/ln/dvmm/vso/download/sentibank.html or got to the foler ANP.

[Disclaimer]

Please refer to our forthcoming paper at MIS Quarterly for details: Pictures that are Worth a Thousand Donations: How Emotions in Project Images Drive the Success of Online Charity Fundraising Campaigns? An Image Design Perspective, Jian-Ren Hou, Jennifer Zhang, and Kunpeng Zhang.