Unsupervised object detection via LWIR/RGB translation

In this work, we present two new methods to overcome the lack of annotated long-wavelength infrared (LWIR) data by exploiting the abundance of similar RGB imagery. We introduce a novel unsupervised adaptation to the cycleGAN architecture for translating non-corresponding LWIR/RGB datasets. Our ultimate goal is high detection rates in the real LWIR imagery using only RGB labelled imagery for training detection algorithms. In our first experiment, we translate LWIR imagery to RGB, allowing us to use an RGB trained detection algorithm. We, thereby remove the need for labelled LWIR imagery for training detection algorithms. Experimental results show that our adaption helps to create synthetic RGB imagery with higher detection rates across two different datasets. We also find that combining the synthetic RGB and real LWIR imagery produces higher F1 scores on the RGB trained detection network. In our second experiment, we translate RGB to LWIR to fine-tune a network for detection in real LWIR imagery. This method produces the highest F1 scores out of the two methods with detection reaching up to 85.6%.