YOLOTrack 1.1

M. Fränzl

Molecular Nanophotonics Group, Peter Debye Institute for Soft Matter Physics, Universität Leipzig, Linnestr. 5, 04103 Leipzig, Germany

[1] M. Fränzl and F. Cichos, Active Particle Feedback Control with a Single‑Shot Detection Convolutional Neural Network, Sci. Rep. 10, 12571 (2020).

Content

• Description
• Repository Contents
• Users Guide
• System Requirements
• Setup Instructions
• License

Description

This is an extension of the YOLOTrack 1.0 to detect oriented bounding boxes.

Fig. 1 YOLOTrack 1.1 detection for a synthetic image of Janus-type particles as observed in a darkfield microscope.

Repository Contents

• YOLOTrack 1.1 Training (Python): Jupyter notebooks and Python modules for synthetic image generation, training and exporting YOLOTrack 1.1 models with Python/Keras using the TensorFlow backend
• TF DLL: DLL for running TensorFlow models on GPU (TF.dll)
• YOLOTrack 1.1 DLL: DLL for decoding the YOLOTrack 1.1 output tensor (YOLOTrack11.dll)
• YOLOTrack 1.1 Inference (LabVIEW): Example for importing and running a trained YOLOTrack 1.1 model in LabVIEW
• YOLOTrack 1.1 Inference (C++): Example for importing and running a trained model with Qt/C++

Users Guide

See YOLOTrack 1.0.

System Requirements

See YOLOTrack 1.0.

Setup Instructions

For setup instructions refer to the README.md in each directory.