ReLIC

Pytorch code for our work “Representation Learning of Image Composition for Aesthetic Evaluation”.

Citation

Lin Zhao, Meimei Shang, Fei Gao*, et al. Representation Learning of Image Composition for Aesthetic Prediction. Computer Vision and Image Understanding (CVIU), vol. 199, 103024, Oct. 2020. [paper]

Framework

Requirements

• pytorch
• torchvision
• tqdm
• requests

Code (folder)

• It contains AVA, CPC, JAS_composition, JAS_aesthetic.
  • AVA: aesthetic prediction on the AVA dataset;
  • CPC: composition prediciotn on the CPC dataset;
  • JAS_composition: composition prediction on the JAS dataset;
  • JAS_aesthetic: aesthetic prediction on the JAS dataset;
• Pretrained models are released in pretrain_model
  • e denotes ReLIC_e
  • u denotes ReLIC_u
  • ReLIC denotes ReLIC
  • ReLIC1 denotes ReLIC+
  • ReLIC2 denotes ReLIC++
• you can change the 'path_to_model_weight' in option.py and run start_check_model in main.py
• if you want to train your own models, please run start_train in main.py

Feel free to ask any questions about coding.

• Meimei Shang, MeimShang@163.com

Data (folder)

• data contains the dataset split of three datasets: AVA, JAS, CPC;
  • [AVA: A Large-Scale Database for Aesthetic Visual Analysis]（http://refbase.cvc.uab.es/files/MMP2012a.pdf）
  • CPC: [The Comparative Photo Composition (CPC) database]（https://www3.cs.stonybrook.edu/~cvl/projects/wei2018goods/VPN_CVPR2018s.html）
  • JAS: [JENAESTHETICS DATASET- A PUBLIC COLLECTION OF PAINTINGS FOR AESTHETICS RESEARCH]（http://www.inf-cv.uni-jena.de/jenaesthetics.html）
• each of them have three files: train.csv, test.csv and val.csv

Results

BibTex

@article{Zhao2020ReLIC,
title = "Representation learning of image composition for aesthetic prediction",
author = "Lin Zhao and Meimei Shang and Fei Gao and Rongsheng Li and Fei Huang and Jun Yu",
journal = "Computer Vision and Image Understanding",
volume = "199",
pages = "103024",
year = "2020",
issn = "1077-3142",
doi = "https://doi.org/10.1016/j.cviu.2020.103024",
}

References (selected)

• Amirshahi, S.A., Hayn-Leichsenring, G.U., Denzler, J., Redies, C., 2014a. Jenaaesthetics dataset URL: http://www.inf-cv.uni-jena.de/en/jenaesthetics.
• Amirshahi, S.A., Hayn-Leichsenring, G.U., Denzler, J., Redies, C., 2014b. Jenaesthetics subjective dataset: analyzing paintings by subjective scores, in: European Conference on Computer Vision, Springer. pp. 3–19.
• Deng, Y., Chen, C.L., Tang, X., 2017. Image aesthetic assessment: An experimental survey. IEEE Signal Processing Magazine 34, 80–106.
• Murray, N., Marchesotti, L., Perronnin, F., 2012. AVA: A large-scale database for aesthetic visual analysis, in: IEEE International Conference on Computer Vision and Pattern Recognition, pp. 2408–2415.
• Wei, Z., Zhang, J., Shen, X., Lin, Z., Mech, R., Hoai, M., Samaras, D., 2018. Good view hunting: Learning photo composition from dense view pairs, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 5437–5446.
• Ma, S., Liu, J., Wen Chen, C., 2017. A-lamp: Adaptive layout-aware multipatch deep convolutional neural network for photo aesthetic assessment , 4535–4544.
• Talebi, H., Milanfar, P., 2018. NIMA: Neural image assessment. IEEE Transactions on Image Processing 27, 3998–4011. doi:10.1109/TIP.2018.2831899.