Figure 6 - available via license: Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International
Content may be subject to copyright.
Source publication
Automatically generating maps from satellite images is an important task. There is a body of literature which tries to address this challenge. We created a more expansive survey of the task by experimenting with different models and adding new loss functions to improve results. We created a database of pairs of satellite images and the correspondin...
Contexts in source publication
Context 1
... as will be seen in the results, the generated map is more accurate in the features it produces since the generator architecture demands a pixel-wise image translation/pixel-wise coloring. For example, if we look at tile number 6 in figure 6, we see that the generated map is more accurate and has more features than the real map when compared with the available satellite image. 8. To the best of our knowledge, ours is the first model to incorporate a reconstruction loss (for pixel-wise accuracy) and a style loss (to reduce high frequency artifacts) in addition to the GAN loss (a feature-wise learnt similarity metric or content loss similar to the ideas presented in Gatys, Ecker, and Bethge (2015) and Larsen, Sønderby, and Winther (2015)) for the task of generating the standard layer of the map from a satellite image. ...
Context 2
... model performed, by far, the best, especially when trained with the reconstruction and style losses. The training curves for one of our runs is shown in figure 5. Some examples of maps generated using this model trained on the CycleGAN dataset can be found in figure 6. More samples can be found in the appendix. ...
Similar publications
Ensemble coding creates compressed representations of a stimulus array. When discriminating the ensemble average against a reference, however, items in the ensemble with feature values closer to the reference are typically weighed stronger. We have recently shown that this "robust averaging" behavior can be explained as a form of efficient coding,...
Automatically generating maps from satellite images is an important task. There is a body of literature which tries to address this challenge. We created a more expansive survey of the task by experimenting with different models and adding new loss functions to improve results. We created a database of pairs of satellite images and the correspondin...
In recent years, there has been a growing interest in the use of Generative Adversarial Networks (GANs). Thanks to their outstanding performance in image translation and generation, they play an increasingly important role in computer vision applications. Most approaches based on GAN focus on proposing task-specific auxiliary modules or loss functi...
Identity-based Deepfake detection methods have the potential to improve the generalization, robustness, and interpretability of the model. However, current identity-based methods either require a reference or can only be used to detect face replacement but not face reenactment. In this paper, we propose a novel Deepfake video detection approach bas...
Myrmica sabuleti ants have a mental number line on which numbers (non-symbolic displayed amounts) are ranked, the smaller on the left and the larger on the right. Here we try to know if the difference between two successive numbers is identically estimated all along this line or is less and less well estimated with increasing number magnitude. Ants...
Citations
... This loss is used as an additional loss function of generator and discriminator to measure the degree of difference between the original graph and the generated graph. GeoGAN (Ganguli et al. 2019) proposed style losses and reconstruction losses to generators and discriminators to force the generated graph to have similar styles and textures to the original. LE-GAN (Fu et al. 2022) added identity invariant loss, and trained the generator and discriminator by extracting feature vectors from the generated graph and the original graph to ensure the identity consistency between the generated graph and the original graph. ...
Magnetic resonance images (MRIs) of different modalities have different reference values for pathological diagnosis. The difficulty of obtaining multimodal MRIs makes it effective to synthesize medical images of missing modalities from existing one. To train a one-for-all synthesizer with limited number of unpaired MRIs, a novel cross-dimensional knowledge-guided generative adversarial network (CKG–GAN) is proposed. In CKG–GAN, a cross-dimensional knowledge transfer network is utilized to measure the perceptual similarity of 2D images (slices of MRIs) from source and synthesized modalities, the knowledge of which is transferred from a pre-trained 3D network without accessing its private training data set. We evaluate the proposed model for three tasks on BraTs2018 and BraTs2021 data sets, using one modality of T1, T2 or Flair to synthesize the other two modalities without changing the content. The results show that compared with the current state-of-the-art methods, our method improves the performance by 2–7%.
