Fig 4 - uploaded by Mehmet yamaç
Content may be subject to copyright.
Proxy images recovered when random sensing is used at different sensing rate For the LS CN N algorithm [15], we learned the sampling matrix for the MNIST dataset, i.e. we get the Ψ and Ψ T matrices in Eq. 4 from the first and second fully connected

Proxy images recovered when random sensing is used at different sensing rate For the LS CN N algorithm [15], we learned the sampling matrix for the MNIST dataset, i.e. we get the Ψ and Ψ T matrices in Eq. 4 from the first and second fully connected

Source publication
Preprint
Full-text available
Compressive Sensing (CS) theory asserts that sparse signal reconstruction is possible from a small number of linear measurements. Although CS enables low-cost linear sampling, it requires non-linear and costly reconstruction. Recent literature works show that compressive image classification is possible in CS domain without reconstruction of the si...

Contexts in source publication

Context 1
... examples of proxy images recovered with the trans- pose of the random Gaussian matrix using Eq. 4 are shown in Figure 4 for four sampling rates. Starting from such a proxy image, we compute CNN features (coefficients of the fully connected last layer) using the Lenet5 model [14]. ...
Context 2
... examples of proxy images recovered with the trans- pose of the random Gaussian matrix using Eq. 4 are shown in Figure 4 for four sampling rates. Starting from such a proxy image, we compute CNN features (coefficients of the fully connected last layer) using the Lenet5 model [14]. ...

Similar publications

Preprint
Full-text available
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...
Technical Report
Full-text available
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...
Article
Full-text available
The diagnosis of small-bowel tumors is challenging due to their low incidence, nonspecific presentation, and limitations of traditional endoscopic techniques. In our study, we examined the utility of the mucosal protrusion angle in differentiating between true submucosal masses and bulges of the small bowel on video capsule endoscopy. We retrospect...