Comparative analysis: Building Extraction Methods

İrem Kömürcü
2 min readMay 9, 2022


I am happy to share that our article “Comparative analysis of deep learning based building extraction methods with the new VHR Istanbul dataset” has just been published in the journal of Expert Systems with Applications. My first academic paper is online! (Q1, IF: 6.954)

More than 60 experiments were conducted by applying state-of-the-art architectures such as U-Net, Unet++, DeepLabv3+, FPN and PSPNet with different pre-trained encoders and hyperparameters. Our experiments showed that Unet++ architecture using SE-ResNeXt101 encoder pre-trained with ImageNet provides the best results with 93.8% IoU on the Istanbul dataset.

Predictions from the test dataset. Raw predictions (a and c), post-processed (b and d).

Our Paper: Comparative analysis of deep learning-based building extraction methods with the new VHR Istanbul dataset:

This link above provides 50 days of free access to the article. Anyone clicking on this link before June 26, 2022, will be taken directly to the final version of your article on ScienceDirect, which they are welcome to read or download.

I created a repo for the paper. You can access the repo:

Original repo:

I wanna share special thanks to Prof. Dr. Elif Sertel and host Dr. Tolga Bakirman

Journal: Expert Systems with Applications

Feel free to reach me! I regularly try to write resource recommendations and technical articles. You can follow my Medium account, if you like the article, you can present your appreciation with claps. Your comments and interaction with me will make me happy.

You can follow and communicate with me on social media, especially Twitter. Thanks!



İrem Kömürcü

Google Developer Expert on Machine Learning | Data Scientist @Deloitte |