Applications of object detection networks in high-power laser systems and experiments

Abstract

The recent advent of deep artificial neural networks has resulted in a dramatic increase in performance for object classification and detection. While pre-trained with everyday objects, we find that a state-of-the-art object detection architecture can very efficiently be fine-tuned to work on a variety of object detection tasks in a high-power laser laboratory. In this manuscript, three exemplary applications are presented. We show that the plasma waves in a laserplasma accelerator can be detected and located on the optical shadowgrams. The plasma wavelength and plasma density are estimated accordingly. Furthermore, we present the detection of all the peaks in an electron energy spectrum of the accelerated electron beam, and the beam charge of each peak is estimated accordingly. Lastly, we demonstrate the detection of optical damage in a high-power laser system. The reliability of the object detector is demonstrated over one thousand laser shots in each application. Our study shows that deep object detection networks are suitable to assist online and offline experiment analysis, even with small training sets. We believe that the presented methodology is adaptable yet robust, and we encourage further applications in Hz-level or KHz-level high-power laser facilities regarding the control and diagnostic tools, especially for those involving image data.

Tango Controls and data pipeline for petawatt laser experiments

“…As an example, Figure 9 shows a plot of the air temperature against the corresponding These developments lay the foundation for more advanced machine learning-based data processing and control techniques [12] by providing stored data, as well as live measurements in a standardized format. For instance, researchers at CALA have been actively working on developing control systems based on Bayesian optimization [18] and have used object detection networks to actively monitor various features or patterns in diagnostics (e.g., optical damage or few-cycle images of laser-driven plasma waves) [19] . As an example, Figure 10 shows objects detected by an object detection network [20] , which was fine-tuned to detect and distinguish features that regularly occur in few-cycle shadowgraphy, such as plasma waves or diffraction patterns from dust on optics or hydrodynamic shocks.…”

Data-driven science and machine learning methods in laser–plasma physics

“…The axial distance between the damage site and the imaging system is obtained numerically by the principle of holographic focusing. More examples for applications of object detection in a high-power laser laboratory have been reported in the work of Lin et al [ 273 ] . In addition to the aforementioned case of optical damages in a high-power laser beamline, the authors fine-tuned the YOLO network for object detection in the few-cycle shadowgraphy of plasma waves and electron beam detection in an electron spectrometer.…”