凌檢測：電視機背板螺絲位置檢測

Abstract

本研究著眼於開發一種創新的電腦視覺演算法，名為"凌檢測"，專門用於精確識別和量測電視機背板上螺絲的位置和尺寸。這一演算法的開發是為了實現比人工檢測更高的速度和效率，尤其是在螺絲位置分佈多樣且精確定位至關重要的情境下。 電視機背板上的螺絲分佈依品牌和型號而異，固定方式亦有所不同，包括表面固定的裸露螺絲和位於機殼深處的隱藏螺絲。因此，「凌檢測」演算法被設計為能夠兼容並準確檢測各種電視機背板上所有螺絲的位置。 研究中使用實際檢測到的螺絲準確率作為評估演算法性能的標準。鑒於後續工序需要將這些位置信息提供給機械臂以便進行電視機的拆解和回收，準確性成為至關重要的因素。任何漏檢的螺絲都可能導致拆解工作的失敗。因此，本研究的實驗目標是利用新開發的人工智能演算法，在保證高準確度的前提下，高效完成螺絲位置的檢測工作。

This study focuses on the development of an innovative computer vision algorithm, named "LingInspect," specifically designed to accurately identify and measure the position and size of screws on the back panels of televisions. The development of this algorithm is intended to achieve greater speed and efficiency than manual inspection, particularly in scenarios where diverse screw placements require precise localization. The distribution of screws on the back panels of televisions varies by brand and model, and the methods of fixation differ as well, including screws that are surface-mounted and exposed, as well as those that are set deeper, below the casing height. Thus, LingInspect algorithm is designed to be compatible with and accurately detect the positions of all screws on various television back panels. In this research, the accuracy of screw detection as measured by the algorithm serves as the standard for assessing its performance. Given that subsequent processes require providing these position data to robotic arms for the disassembly and recycling of televisions, accuracy is of paramount importance. Any missed screws could result in the failure of the disassembly process. Therefore, the experimental objective of this study is to utilize the newly developed artificial intelligence algorithm to efficiently complete screw position detection with high accuracy and speed.