Video annotation is a critical and time-consuming task in computer vision research and applications. This paper presents a novel annotation pipeline that uses pre-extracted features and dimensionality reduction to accelerate the temporal video annotation process. Our approach uses Hierarchical Stochastic Neighbor Embedding (HSNE) to create a multi-scale representation of video features, allowing annotators to efficiently explore and label large video datasets. We demonstrate significant improvements in annotation effort compared to traditional linear methods, achieving more than a 10x reduction in clicks required for annotating over 12 hours of video. Our experiments on multiple datasets show the effectiveness and robustness of our pipeline across various scenarios. Moreover, we investigate the optimal configuration of HSNE parameters for different datasets. Our work provides a promising direction for scaling up video annotation efforts in the era of video understanding.

Scheduling is required in almost every industry and when done well it can bring a lot of revenue. Flexibility is often forgotten when creating the initial schedules. Therefore, in case of an unexpected delay, the whole schedule has to suffer. In this paper, we consider a re-entrant flow shop with sequence-dependent setup times and relative due dates for our industrial partner, which specialises in industrial printers. Then, we perform a robustness analysis on real schedules from the industry, which can be extended to any system represented as a flow shop with relative due dates. We find how much time a schedule with relative due dates has before it becomes infeasible. We continue by empirically creating a new robustness measure and comparing it with state-of-the-art techniques. The experiments confirm that this measure can be useful in creating robust initial schedules for re-entrant flow shops with added idle time that has a minimal effect on the total duration of the solution.