超多分类的Softmax 2014年CVPR两篇超多分类的人脸识别论文:DeepFace和DeepID DeepFace Taigman Y, Yang M, Ranzato M A, et al. Deepface: Closing the gap to humanlevel performance in face verification [C]// CVPR, 2014. 4.4M训练集,训练6层CNN + 4096特征映射 + 4030类Softmax,综合如3D Aligement, model ensembel等技术,在LFW上达到97.35%。 DeepID Sun Y, Wang X, Tang X. Deep learning face representation fro...
题目 Given two sorted integer arrays nums1 and nums2, merge nums2 into nums1 as one sorted array. Note: The number of elements initialized in nums1 and nums2 are m and n respectively. You may assume that nums1 has enough space (size that is equal to m + n) to hold additional elements from nums2. Example: [代码] Constraints: 10^9 <= nums1[i], nums2[i] <...
SPP (spatial pyramid pooling layer) SPP applies a slightly different strategy in detecting objects of different scales. It replaces the last pooling layer (after the last convolutional layer) with a spatial pyramid pooling layer. The feature maps are spatially divided into m×m bins with m, say, equals 1, 2, and 4 respectively. Then a maximum pool i...