In recent years, big data became a hard challenge. Analyzing big data needs a lot of speed precision combination. In this article, we describe a deep learning-based method to deal with big data with a focus on precision and speed. In our case, the data are images that are the hardest type of data to manipulate because of their complex structure that needs a lot of computation power. Besides, we will solve a hard task on images, which is object detection and identification. Thus, every object in the image will be localized and classified according to the range of classes provided by the training data set. To solve this challenge, we propose an approach based on a deep convolutional neural network (CNN). Moreover, CNN is the most used deep learning model in computer vision tasks such as image classification and object recognition because of its power in self-features extraction and provides useful techniques in the prediction of decision-making. Our approach outperforms state-of-the-art models such as R-CNN, Fast R-CNN, Faster R-CNN, and YOLO (you only look once), with 77% of mean average precision on the Pascal_voc 2007 testing data set and a speed of 16.54 FPS using an Nvidia Geforce GTX 960 GPGPU.

中文翻译：

---

在视觉大规模数据中执行目标检测和识别的卷积神经网络

近年来，大数据成为一项艰巨的挑战。分析大数据需要很多速度精度的组合。在本文中，我们描述了一种基于深度学习的方法来处理大数据，重点是精度和速度。在我们的例子中，数据是最难处理的数据类型，因为它们的结构复杂，需要大量的计算能力。此外，我们将解决图像上的一项艰巨任务，即对象检测和识别。因此，图像中的每个对象都将根据训练数据集提供的类别范围进行定位和分类。为了解决这一挑战，我们提出了一种基于深度卷积神经网络 (CNN) 的方法。而且，CNN 是计算机视觉任务（如图像分类和对象识别）中最常用的深度学习模型，因为它具有强大的自我特征提取能力，并为决策预测提供了有用的技术。我们的方法优于最先进的模型，例如 R-CNN、Fast R-CNN、Faster R-CNN 和 YOLO（您只看一次），在 Pascal_voc 2007 测试数据集上的平均精度为 77%使用 Nvidia Geforce GTX 960 GPGPU 的速度为 16.54 FPS。