TESTING THE PERFORMANCE FOR UNRELATED COLOUR APPEARANCE MODELS

Most colour appearance models (CAMs) have been developed to predict the colour appearance only for related colours in different viewing conditions. More recently, CAMs to predict for unrelated colours have been proposed due to the availability of the experimental data. This paper investigates the performance of the three promising colour appearance models for unrelated colours, i.e. CAM15u, CAMFu and CAM20u using an experimental dataset carefully accumulated by Fu et al. in term of CV values. The results showed that the latest CAM20u outperformed the other models. CAMFu, as the oldest model, whose performance was much worse than other two models in predicting brightness and colourfulness, and performed well in predicting hue composition, while CAM15u gave a moderate performance in predicting brightness and colourfulness, but performed slightest worse than the others for hue composition.

Real-Time Mask Recognition

While a variety of image processing studies have been performed to quantify the potential performance of neural network-based models using high-quality still images, relatively few studies seek to apply those models to a real-time operational context. This paper seeks to extend prior work in neural-network-based mask detection algorithms to a real-time, low-power deployable context that is conducive to immediate installation and use. Particularly relevant in the COVID-19 era with varying rules on mask mandates, this work applies two neural network models to inference of mask detection in both live (mobile) and recorded scenarios. Furthermore, an experimental dataset was collected where individuals were encouraged to use presentation attacks against the algorithm to quantify how perturbations negatively impact model performance. The results from evaluation on the experimental dataset are further investigated to identify the degradation caused by poor lighting and image quality, as well as to test for biases within certain demographics such as gender and ethnicity. In aggregate, this work validates the immediate feasibility of a low-power and low-cost real-time mask recognition system.

Ext4 and XFS File System Forensic Framework Based on TSK

Recently, the number of Internet of Things (IoT) devices, such as artificial intelligence (AI) speakers and smartwatches, using a Linux-based file system has increased. Moreover, these devices are connected to the Internet and generate vast amounts of data. To efficiently manage these generated data and improve the processing speed, the function is improved by updating the file system version or using new file systems, such as an Extended File System (XFS), B-tree file system (Btrfs), or Flash-Friendly File System (F2FS). However, in the process of updating the existing file system, the metadata structure may be changed or the analysis of the newly released file system may be insufficient, making it impossible for existing commercial tools to extract and restore deleted files. In an actual forensic investigation, when deleted files become unrecoverable, important clues may be missed, making it difficult to identify the culprit. Accordingly, a framework for extracting and recovering files based on The Sleuth Kit (TSK) is proposed by deriving the metadata changed in Ext4 file system journal checksum v3 and XFS file system v5. Thereafter, by comparing the accuracy and recovery rate of the proposed framework with existing commercial tools using the experimental dataset, we conclude that sustained research on file systems should be conducted from the perspective of forensics.