Publications

Resources / Publications

Dragonfly as a platform for easy image-based deep learning applications

Ruslana Makovetsky (1), Nicolas Piche (1), Mike Marsh (2)

Microscopy and Microanalysis, 24(S1), 01 August 2018:532-533. DOI: 10.1017/S143192761800315X

---

Keywords

Deep learning; Image processing; Dragonfly; Image segmentation; Image visualization; Image analysis

---

Abstract

Deep learning (DL) refers to a class of model-based computational techniques for processing data and solving problems, often with results that deliver higher accuracy compared to human experts executing the same tasks. Adoption of DL solutions has accelerated in recent years in part because of success in domains that have been otherwise intractable with traditional algorithms and in part because of the rapidly growing speed of high-performance DL engines. Here, we apply the popular DL engine TensorFlow [1] to image processing tasks and make it user-friendly for microscopists and other image scientists to apply convolutional neural networks (CNNs) in our Dragonfly software platform. We discuss applications in image filtering and image segmentation…

---

Author Affiliation

(1) Object Research Systems, Montreal, Canada
(2) Object Research Systems, Denver, CO, USA

---

Return to Publications list.