Automated complaint analysis is vital for generating critical insights, which in turn enhance customer satisfaction, product quality, and overall business performance. Nevertheless, conventional methods frequently fail to capture the nuances of aspect-level complaints and inadequately utilize external knowledge, thus creating a gap in effective complaint detection and analysis. In response to this issue, we proactively explore the role of external knowledge and multimodality in this domain. This leads to the development of MGasD (Multimodal Generative framework for aspect-based complaint and sentiment Detection), a multimodal knowledge-infused unified framework. MGasD diverges from traditional methods by reframing the complaint detection problem as a multimodal text-to-text generation task. Significantly, our research includes the development of a novel aspect-level dataset. Annotated for both complaint and sentiment categories across diverse domains such as books, electronics, edibles, fashion, and miscellaneous, this dataset provides a comprehensive platform for the concurrent study of complaints and sentiment. This resource facilitates a more robust understanding of consumer feedback. Our proposed methodology establishes a benchmark performance in the novel aspect-based complaint and sentiment detection tasks based on extensive evaluation. We also demonstrate that our model consistently outperforms all other baselines and state-of-the-art models in both full and few-shot settings (The dataset and code are available at:https://github.com/appy1608/CIKM2023).