Recent advances in text-to-image models have enabled a new era of creative and controllable image generation. However, generating compositional scenes with multiple subjects and attributes remains a significant challenge. To enhance user control over subject placement, several layout-guided methods have been proposed. However, these methods face numerous challenges, particularly in compositional scenes. Unintended subjects often appear outside the layouts, generated images can be out-of-distribution and contain unnatural artifacts, or attributes bleed across subjects, leading to incorrect visual outputs. In this work, we propose MALeR, a method that addresses each of these challenges. Given a text prompt and corresponding layouts, our method prevents subjects from appearing outside the given layouts while being in-distribution. Additionally, we propose a masked, attribute-aware binding mechanism that prevents attribute leakage, enabling accurate rendering of subjects with multiple attributes, even in complex compositional scenes. Qualitative and quantitative evaluation demonstrates that our method achieves superior performance in compositional accuracy, generation consistency, and attribute binding compared to previous work. MALeR is particularly adept at generating images of scenes with multiple subjects and multiple attributes per subject.

Electric vehicles (EVs) are becoming the promising technology for achieving a sustainable transport sector, because of their ability to zero carbon emissions, high efficiency, and flexibility in grid operation & integration. The e-Powertrain (ePWT) powers EVs and eliminates need for an internal combustion engine from vehicles due to its lightweight & compactness that produces extremely low vibrations and generates instant torque for e-mobility. The e-Powertrain consists of the main units: power delivery module (battery pack & Battery Management System (BMS), charger, DC-DC converters), high-power electric motor, inverter, and reduction drive. BMS is well known as the brain of all ePWT systems.Model-based design (MBD) approach adopted in development of e-powertrain software is presented in this research paper. MBD approach is preferred here for automotive software development to save time, to ensure reliability and cost saving. The MBD process shifts the emphasis on functions development from requirements rather than the manual code for automotive software development. A model, a diagrammatic representation of requirements, is created and tested at multiple stages using component testing & integration testing strategies before the code generation using tools like MATLAB & Simulink which minimizes the chances of errors in software implementation. This paper demonstrates the features and benefits of system design using MBD including testing/evaluation and documentation which result in reduced time and cost for development. In this paper, a specific MBD process is used and demonstrated for BMS software development for xEVs (EVs & HEVs) which manages EV battery, by protecting it to operate beyond its safe limits and provides safety to the system and user with run time monitoring of battery status for any critical hazarder conditions. Further, it illustrates main steps of the modeling process & show its implementation, functional simulation results and reliability. State-off-the-art algorithms are used for computation in BMS for sake of simplicity which can easily be updated for other empirical & machine learning algorithms.

We introduce multimodal story summarization by lever-aging TV episode recaps - short video sequences interweaving key story moments from previous episodes to bring viewers up to speed. We propose PlotSnap, a dataset featuring two crime thriller TV shows with rich recaps and long episodes of 40 minutes. Story summarization labels are unlocked by matching recap shots to corresponding sub-stories in the episode. We propose a hierarchical model TaleSumm that processes entire episodes by creating compact shot and dialog representations, and predicts importance scores for each video shot and dialog utterance by enabling interactions between local story groups. Unlike traditional summarization, our method extracts multiple plot points from long videos. We present a thorough evaluation on story summarization, including promising cross-series generalization. TaleSumm also shows good results on classic video summarization benchmarks.

Movie story analysis requires understanding characters' emotions and mental states. Towards this goal, we formulate emotion understanding as predicting a diverse and multi-label set of emotions at the level of a movie scene and for each character. We propose EmoTx, a multimodal Transformer-based architecture that ingests videos, multiple characters, and dialog utterances to make joint predictions. By leveraging annotations from the MovieGraphs dataset [72], we aim to predict classic emotions (e.g. happy, angry) and other mental states (e.g. honest, helpful). We conduct experiments on the most frequently occurring 10 and 25 labels, and a mapping that clusters 181 labels to 26. Ablation studies and comparison against adapted state-of-the-art emotion recognition approaches shows the effectiveness of EmoTx. Analyzing EmoTx's self-attention scores reveals that expressive emotions often look at character tokens while other mental states rely on video and dialog cues.