A Framework to Estimate the Key Point Within an Object Based on a Deep Learning Object Detection

Abstract

Automatic identification of key points within objects is crucial in various application domains. This paper presents a novel framework for accurately estimating the key point within an object by leveraging deep neural network-based object detection. The proposed framework is built upon a training dataset annotated with four non-overlapping bounding boxes, one of which shares a coordinate with the key point. These bounding boxes collectively cover the entire object, enabling automatic annotation if region annotations around the key point exist. The trained object detector is then utilized to generate detection results, which are subsequently post-processed to estimate the key point. To validate the effectiveness of the framework, experiments were conducted using two distinct datasets: cross-sectional images of a parawood log and pupil images. The experimental results demonstrate that our proposed framework surpasses previously proposed approaches in terms of precision, recall, F1-score, and other domain-specific metrics. The improvement in performance can be attributed to the unique annotation strategy and the fusion of object detection and key point estimation within a unified deep learning framework. The contribution of this study lies in introducing a novel framework for closely estimating key points within objects based on deep neural network-based object detection. By leveraging annotated training data and post-processing techniques, our approach achieves superior performance compared to existing methods. This work fills a critical gap in the field by integrating object detection and key point estimation, which has received limited attention in previous research. Our framework provides valuable insights and advancements in key point estimation techniques, offering potential applications in precise object analysis and understanding. Doi: 10.28991/HIJ-2023-04-01-08 Full Text: PDF

A Fine-Tuned RetinaNet for Real-Time Lettuce Detection

“…Annotations are the information associated with each object in the images or videos. Annotations include bounding box coordinates surrounding the thing, class labels, and other attributes such as pose or object landmarks [27]. Annotations are essential for training object recognition models to recognize and learn the characteristics of target objects.…”

Traffic Scene Analysis and Classification using Deep Learning

“…Kurdthongmee et al (10) studied the YOLOv7 method which was proposed for detection in crosssectional images of a parawood log and pupil datasets. This study aims to identify critical points in these images.…”

Process modeling and simulation of Gmelina arborea (GmW) and Mansonia altissima (MaW) wood drying

“…Because of its unique specific strength and aesthetic quality, it is also useful for artistic purposes [ 4 , 5 ]. Following a surge in infrastructure development, there is an increasing trend in wood harvesting, and it has been well established that trees have a high moisture content that contributes to their functional make-up [ 6 , 7 ]; Argo & Ubaidilla, 2020; [ 8 ]. As a result, freshly cut lumber has high moisture content, as expected.…”

“…The lumbering industry has suffered significant economic losses and waste as a result of a lack of commensurate effort toward effective post-harvest processing. To address these concerns, the shelf life and suitability of harvested timbers (for many of their various applications) should be increased by removing excess moisture through seasoning [ 8 , 9 ]. Drying is an important part of wood seasoning and preparing it for further use in the construction industry.…”