Deep neural networks (DNNs) achieve state-of-the-art performance in video anomaly detection. However, the usage of DNNs is limited in practice due to their computational overhead, generally requiring significant resources and specialized hardware. Further, despite recent progress, current evaluation criteria of video anomaly detection algorithms are flawed, preventing meaningful comparisons among algorithms. In response to these challenges, we propose (1) a compression-based technique referred to as Spatio-Temporal N-Gram Prediction by Partial Matching (STNG PPM) and (2) simple modifications to current evaluation criteria for improved interpretation and broader applicability across algorithms. STNG PMM does not require specialized hardware, has few parameters to tune, and is competitive with DNNs on multiple benchmark data sets in video anomaly detection.
The primary purpose of this document is to outline the progress made on the LDRD titled “Identifying and Explaining Anomalous Activity in Surveillance Video with Compression Algorithms” in FY22 and FY23. In this LDRD, we explored the usage of compression-based analytics to identify anomalous activity in video. We developed a novel algorithm, Spatio-Temporal N-Gram PPM (STNG PPM) that accounts for spatially and temporally aware anomalies in video. We extracted features using motions vectors from video as well as operating on the raw features. STNG PPM is comparable to many deep learning approaches but does not require specialized hardware (GPUs) to run efficiently. We also examine the evaluation metrics and propose novel measures addressing faults in the current evaluation measures.
A series of outdoor shots were conducted at the HERMES III facility in November 2016. There were several goals associated with these experiments, one of which is an improved understanding of the courtyard radiation environment. Previous work had developed parametric fits to the spatial and temporal dose rate in the area of interest. This work explores the inter-shot variation of the dose in the courtyard, updated fit parameters, and an improved dose rate model which better captures high frequency content. The parametric fit for the spatial profile is found to be adequate in the far-field, however near-field radiation dose is still not well-understood.
This report describes data collection and analysis of solar photovoltaic (PV) equipment events, which consist of faults and fa ilures that occur during the normal operation of a distributed PV system or PV power plant. We present summary statistics from locations w here maintenance data is being collected at various intervals, as well as reliability statistics gathered from that da ta, consisting of fault/failure distributions and repair distributions for a wide range of PV equipment types.
This user manual is intended to provide instructions to volunteer beta testers on how to use Sandia National Laboratories (SNL) PV Reliability Performance Model (PV-RPM) features in the National Renewable Energy Laboratory (NREL) System Advisor Model (SAM) version 2017.1.17 r4 (NREL, 2017). This new feature is provided in SAM to allow users with reliability data the ability to develop and run scenarios where PV performance and costs are impacted from components that can fail stochastically. This is intended to be an advanced user feature as it requires knowledge and data regarding different PV component failure modes. It also relies heavily on the SAM LK scripting language, which is not utilized by a majority of SAM users. NREL has published a SAM LK users guide (Dobos, 2017) and has multiple online help topics and videos to get users familiar with the scripting language and what it can do. This user instruction manual will provide some background on how data collected from a PV system can be used as inputs in the PV-RPM model, which will give data owners the ability to develop their own reliability and repair distributions outside of the example provided here.