Connecting multiple machine learning models into a pipeline is effective for handling complex problems. By breaking down the problem into steps, each tackled by a specific component model of the pipeline, the overall solution can be made accurate and explainable. This paper describes an enhancement of object detection based on this multi-step concept, where a post-processing step called the calibration model is introduced. The calibration model consists of a convolutional neural network, and utilizes rich contextual information based on the domain knowledge of the input. Improvements of object detection performance by 0.8-1.9 in average precision metric over existing object detectors have been observed using the new model.
翻译:将多台机器学习模型连接到管道中,对于处理复杂问题十分有效。通过将问题分解为步骤,每个步骤都由管道的具体部件模型处理,可以准确和解释整体解决办法。本文件描述了基于这一多步概念的物体探测增强,在此概念的基础上,引入了一个后处理步骤,称为校准模型。校准模型包括一个演进神经网络,并利用基于输入领域知识的丰富背景信息。使用新的模型,观测到物体探测性能以0.8-1.9的平均精确度衡量标准比现有物体探测器提高了0.8-1.9。