基于深度学习的图片中商品参数识别方法

计算机计算性能的提升使得深度学习成为了可能。作为计算机视觉领域的重要发展方向之一的目标检测也开始结合深度学习方法并广泛应用于各行各业。受限于网络的复杂度和检测算法的设计，目标检测的速度和精度成为一个trade-off。目前电商领域的飞速发展产生了大量包含商品参数的图片，使用传统方法难以有效地提取出图片中的商品参数信息。针对这一问题，本文提出了一种将深度学习检测算法和传统OCR技术相结合的方法，在保证了识别速度的同时大大提升了识别的精度。本文研究的问题包括检测模型、针对特定数据训练、图片预处理以及文字识别等。本文首先比较了现有的目标检测算法，权衡其优缺点，然后使用YOLO模型完成检测任务，并针对YOLO模型中存在的不足进行了一定的改进和优化，得到了一个专用于检测图片中商品参数的目标检测模型，最后使用tesseract完成文字提取任务。在将整个流程结合到一起后，我们的系统不仅有着较好的识别精度，而且是高效和健壮的。本文最后还讨论了优势和不足之处，并指出了未来工作的方向。

Deep Learning for Parameter Recognition in Commodity Images

The improvements of computer computing performance make deep learning possible. As one of the important research directions in the field of computer vision, object detection has combined with deep learning methods and is widely used in all walks of life. Limited by the complexity of the network and the design of the detection algorithm, the speed and precision of the object detection becomes a trade-off. At present, the rapid development of electric business field has produced a large number of pictures containing the product parameters. The traditional method is hard to extract the information of the product parameters in the picture. This paper presents a method of combining the deep learning detection algorithm with the traditional OCR technology, which ensures the detection speed and at the same time greatly improves the accuracy of recognition. This paper focuses the following prolems:the detection model, training for specific data, image preprocessing and character recognition. This paper first compares the existing object detection algorithms and weighs their advantages and disadvantages. The YOLO model is used to do the detection work, and we propose some improvements to overcome the shortcomings in the YOLO model. We get an object detection model which is designed to detect the product parameters in images. Finally, we use tesseract to do the character recognition work. The experimental resuts show that our system is efficient and effiective in parameter recognition. At the end of this paper, we discuss the innovation and disadvantage of our method.