Deep learning strategies for accurate identification from facial composite images (E2ID)

“世界各地的警察部队经常使用犯罪嫌疑人的面部合成图像(俗称照相或照片)来协助他们的刑事调查。然而，面部合成作为刑事调查工具的有效性受到目前无法准确和快速地在人群中搜索与合成图像的潜在匹配的严重限制。现有的商用人脸识别系统不能很好地解决这一问题，并且在人脸合成图像上的性能非常差。在这个项目中，我们提出了一种全新的方法来实现人脸合成图像与警察可疑数据库的快速准确匹配。我们将使用人工智能和先进的图像处理方法来生成世界上最大的面部合成图像库。然后，我们将利用这一资源来开发神经(深度学习)程序，成功地映射出面部合成图像识别中隐含的人类认知过程。通过这种方式，机器行为将首次被定制，以类似于人类的方式实现复合人脸识别。预期的发展将促使国际警察部队改进调查程序，从而更好地结案并显著提高效率。一个成功的项目将为将这项技术在世界各地的监管市场中进行商业开发铺平道路。“

"Facial composite images of criminal suspects (commonly known as PhotoFITs or EFITs) are routinely used by police forces to assist their criminal investigations throughout the world. However, the effectiveness of facial composites as tools of criminal investigation is severely limited by the current inability to accurately and rapidly search a population for potential matches to a composite image. Existing commercial face recognition systems do not adequately address this problem and perform very poorly on facial composite images.In this project, we propose a radical, new approach to achieving fast and accurate matching of facial composite images to police suspect databases. We will use methods of artificial intelligence and advanced image processing to generate the world's largest repository of facial composite images. We will then exploit this resource to develop neural (deep learning) procedures that successfully map the human cognitive processes implicit in the recognition of facial composite images. In this way, machine behaviour will be tailored for the first time to achieve composite face recognition in a way similar to human-beings.The expected developments will give rise to improved investigative procedures for international police forces leading to greater case closure and significant efficiency increases. A successful project will pave the way to commercially exploiting this technology in policing markets all over the world."