Kundu, Suman

Automating the detection of fruits and vegetables using computer vision is essential for modernizing agriculture, improving efficiency, ensuring food quality, and contributing to sustainable and technologically advanced farming practices. This paper presents an end-to-end pipeline for detecting and localizing fruits and vegetables in real-world scenarios. To achieve this, a dataset named FRUVEG67 was curated that includes images of 67 classes of fruits and vegetables captured in unconstrained scenarios, with only a few manually annotated samples per class. A semi-supervised data annotation algorithm (SSDA) was developed that generates bounding boxes for objects to label the remaining nonannotated images. For detection, Fruit and Vegetable Detection Network (FVDNet) was proposed, an ensemble version of YOLOv8n featuring three distinct grid configurations. In addition, an averaging approach for the prediction of the bounding box and a voting mechanism for the prediction of the classes was implemented. Finally Jensen–Shannon Divergence (JSD) in conjunction with focal loss was integrated as the overall loss function for better detection of smaller objects. Experimental results highlight the superiority of FVDNet compared to recent versions of YOLO, showcasing remarkable improvements in detection and localization performance. An impressive mean average precision (mAP) score of 0.82 across all classes was achieved. Furthermore, the efficacy of FVDNet on open-category refrigerator images were evaluated, where it demonstrates promising results.

The problem of query-focused multi-document summarization (QMDS) is to generate a summary from multiple source documents on identical/similar topics based on the query submitted by the users. This article provides a systematic review of the literature of QMDS. The research works are classified into six major categories based on the summarization methodologies used. Different techniques used for finding query-relevant summaries for different algorithms under each of the six major groups are reported. Further, 17 evaluation metrics used for evaluating algorithms for text summaries against the human-curated summaries are compiled here in this article. Extensive experiments are performed on eight different datasets. Comparative results of nine methodologies, each representing one of the six different groups, are presented. Seven different evaluation metrics are used in the comparative study. It is observed that DL- and ML-based QMDS methods perform. better in comparison to the other methods.

Accurate action recognition in cooking videos holds significant importance for various applications, such as recipe recommendation systems, dietary monitoring, and interactive cooking tutorials. In this research, we introduce the CookingINWild dataset, a novel collection of diverse Indian cuisine cooking videos sourced from YouTube, capturing real-world, uncontrolled scenarios. We tested our dataset with advanced action recognition models that usually work well. Surprisingly, they didn't do as great on our dataset because Indian cooking actions are complex and our videos aren't controlled. This shows we need special methods for these challenges. Our work connects controlled and real-world situations, making room for better action recognition in Indian cooking videos. The CookingINWild dataset represents a valuable resource for exploring Indian cooking techniques, encouraging the researchers to address these distinctive challenges and enhance action recognition in Indian cooking videos.