This architecture can model local pairwise feature interactions in a translationally invariant manner which is particularly useful for fine-grained categorization.
We present experiments with bilinear models where the feature extractors are based on convolutional neural networks.
Results show that the architecture compares favorably to the existing state of the art on a number of fine-grained datasets while being substantially simpler and easier to train.
Moreover, the architecture can be easily trained end-to-end unlike these texture descriptions leading to significant improvements in performance.
Significant accuracy gains in semantic segmentation have recently been obtained through the use of convolutional networks
Convolutional network architectures that had originally been developed for image classification can be successfully repurposed for dense prediction
The relentless success of deep learning techniques in various high-level computer vision tasks – in particular, supervised approaches such as Convolutional Neural Networks (CNNs) for image classification or object detection – motivated researchers to explore the capabilities of such networks for pixel-level labelling problems like semantic segmentation.
It is designed to be efficient both in terms of memory and computational time during inference.
We present a novel and practical … architecture for …
Sth. has a wide array of applications ranging from … to …
Sth. has seen huge success in …
Sth. is shown both qualitatively and quantitatively to…