使用pytorch实现YOLO（v3）。
How to implement a YOLO (v3) object detector from scratch in PyTorch: Part 1
How to implement a YOLO (v3) object detector from scratch in PyTorch: Part 2
使用人脸数据集实现人脸检测（face detection）。
----------------------------------------------------------------
Layer (type) Output Shape Param #
================================================================
Conv2d-1 [-1, 32, 416, 416] 864
BatchNorm2d-2 [-1, 32, 416, 416] 64
LeakyReLU-3 [-1, 32, 416, 416] 0
Conv2d-4 [-1, 64, 208, 208] 18,432
BatchNorm2d-5 [-1, 64, 208, 208] 128
LeakyReLU-6 [-1, 64, 208, 208] 0
Conv2d-7 [-1, 32, 208, 208] 2,048
BatchNorm2d-8 [-1, 32, 208, 208] 64
LeakyReLU-9 [-1, 32, 208, 208] 0
Conv2d-10 [-1, 64, 208, 208] 18,432
BatchNorm2d-11 [-1, 64, 208, 208] 128
LeakyReLU-12 [-1, 64, 208, 208] 0
Conv2d-13 [-1, 128, 104, 104] 73,728
BatchNorm2d-14 [-1, 128, 104, 104] 256
LeakyReLU-15 [-1, 128, 104, 104] 0
Conv2d-16 [-1, 64, 104, 104] 8,192
BatchNorm2d-17 [-1, 64, 104, 104] 128
LeakyReLU-18 [-1, 64, 104, 104] 0
Conv2d-19 [-1, 128, 104, 104] 73,728
BatchNorm2d-20 [-1, 128, 104, 104] 256
LeakyReLU-21 [-1, 128, 104, 104] 0
Conv2d-22 [-1, 64, 104, 104] 8,192
BatchNorm2d-23 [-1, 64, 104, 104] 128
LeakyReLU-24 [-1, 64, 104, 104] 0
Conv2d-25 [-1, 128, 104, 104] 73,728
BatchNorm2d-26 [-1, 128, 104, 104] 256
LeakyReLU-27 [-1, 128, 104, 104] 0
Conv2d-28 [-1, 256, 52, 52] 294,912
BatchNorm2d-29 [-1, 256, 52, 52] 512
LeakyReLU-30 [-1, 256, 52, 52] 0
Conv2d-31 [-1, 128, 52, 52] 32,768
BatchNorm2d-32 [-1, 128, 52, 52] 256
LeakyReLU-33 [-1, 128, 52, 52] 0
Conv2d-34 [-1, 256, 52, 52] 294,912
BatchNorm2d-35 [-1, 256, 52, 52] 512
LeakyReLU-36 [-1, 256, 52, 52] 0
Conv2d-37 [-1, 128, 52, 52] 32,768
BatchNorm2d-38 [-1, 128, 52, 52] 256
LeakyReLU-39 [-1, 128, 52, 52] 0
Conv2d-40 [-1, 256, 52, 52] 294,912
BatchNorm2d-41 [-1, 256, 52, 52] 512
LeakyReLU-42 [-1, 256, 52, 52] 0
Conv2d-43 [-1, 128, 52, 52] 32,768
BatchNorm2d-44 [-1, 128, 52, 52] 256
LeakyReLU-45 [-1, 128, 52, 52] 0
Conv2d-46 [-1, 256, 52, 52] 294,912
BatchNorm2d-47 [-1, 256, 52, 52] 512
LeakyReLU-48 [-1, 256, 52, 52] 0
Conv2d-49 [-1, 128, 52, 52] 32,768
BatchNorm2d-50 [-1, 128, 52, 52] 256
LeakyReLU-51 [-1, 128, 52, 52] 0
Conv2d-52 [-1, 256, 52, 52] 294,912
BatchNorm2d-53 [-1, 256, 52, 52] 512
LeakyReLU-54 [-1, 256, 52, 52] 0
Conv2d-55 [-1, 128, 52, 52] 32,768
BatchNorm2d-56 [-1, 128, 52, 52] 256
LeakyReLU-57 [-1, 128, 52, 52] 0
Conv2d-58 [-1, 256, 52, 52] 294,912
BatchNorm2d-59 [-1, 256, 52, 52] 512
LeakyReLU-60 [-1, 256, 52, 52] 0
Conv2d-61 [-1, 128, 52, 52] 32,768
BatchNorm2d-62 [-1, 128, 52, 52] 256
LeakyReLU-63 [-1, 128, 52, 52] 0
Conv2d-64 [-1, 256, 52, 52] 294,912
BatchNorm2d-65 [-1, 256, 52, 52] 512
LeakyReLU-66 [-1, 256, 52, 52] 0
Conv2d-67 [-1, 128, 52, 52] 32,768
BatchNorm2d-68 [-1, 128, 52, 52] 256
LeakyReLU-69 [-1, 128, 52, 52] 0
Conv2d-70 [-1, 256, 52, 52] 294,912
BatchNorm2d-71 [-1, 256, 52, 52] 512
LeakyReLU-72 [-1, 256, 52, 52] 0
Conv2d-73 [-1, 128, 52, 52] 32,768
BatchNorm2d-74 [-1, 128, 52, 52] 256
LeakyReLU-75 [-1, 128, 52, 52] 0
Conv2d-76 [-1, 256, 52, 52] 294,912
BatchNorm2d-77 [-1, 256, 52, 52] 512
LeakyReLU-78 [-1, 256, 52, 52] 0
Conv2d-79 [-1, 512, 26, 26] 1,179,648
BatchNorm2d-80 [-1, 512, 26, 26] 1,024
LeakyReLU-81 [-1, 512, 26, 26] 0
Conv2d-82 [-1, 256, 26, 26] 131,072
BatchNorm2d-83 [-1, 256, 26, 26] 512
LeakyReLU-84 [-1, 256, 26, 26] 0
Conv2d-85 [-1, 512, 26, 26] 1,179,648
BatchNorm2d-86 [-1, 512, 26, 26] 1,024
LeakyReLU-87 [-1, 512, 26, 26] 0
Conv2d-88 [-1, 256, 26, 26] 131,072
BatchNorm2d-89 [-1, 256, 26, 26] 512
LeakyReLU-90 [-1, 256, 26, 26] 0
Conv2d-91 [-1, 512, 26, 26] 1,179,648
BatchNorm2d-92 [-1, 512, 26, 26] 1,024
LeakyReLU-93 [-1, 512, 26, 26] 0
Conv2d-94 [-1, 256, 26, 26] 131,072
BatchNorm2d-95 [-1, 256, 26, 26] 512
LeakyReLU-96 [-1, 256, 26, 26] 0
Conv2d-97 [-1, 512, 26, 26] 1,179,648
BatchNorm2d-98 [-1, 512, 26, 26] 1,024
LeakyReLU-99 [-1, 512, 26, 26] 0
Conv2d-100 [-1, 256, 26, 26] 131,072
BatchNorm2d-101 [-1, 256, 26, 26] 512
LeakyReLU-102 [-1, 256, 26, 26] 0
Conv2d-103 [-1, 512, 26, 26] 1,179,648
BatchNorm2d-104 [-1, 512, 26, 26] 1,024
LeakyReLU-105 [-1, 512, 26, 26] 0
Conv2d-106 [-1, 256, 26, 26] 131,072
BatchNorm2d-107 [-1, 256, 26, 26] 512
LeakyReLU-108 [-1, 256, 26, 26] 0
Conv2d-109 [-1, 512, 26, 26] 1,179,648
BatchNorm2d-110 [-1, 512, 26, 26] 1,024
LeakyReLU-111 [-1, 512, 26, 26] 0
Conv2d-112 [-1, 256, 26, 26] 131,072
BatchNorm2d-113 [-1, 256, 26, 26] 512
LeakyReLU-114 [-1, 256, 26, 26] 0
Conv2d-115 [-1, 512, 26, 26] 1,179,648
BatchNorm2d-116 [-1, 512, 26, 26] 1,024
LeakyReLU-117 [-1, 512, 26, 26] 0
Conv2d-118 [-1, 256, 26, 26] 131,072
BatchNorm2d-119 [-1, 256, 26, 26] 512
LeakyReLU-120 [-1, 256, 26, 26] 0
Conv2d-121 [-1, 512, 26, 26] 1,179,648
BatchNorm2d-122 [-1, 512, 26, 26] 1,024
LeakyReLU-123 [-1, 512, 26, 26] 0
Conv2d-124 [-1, 256, 26, 26] 131,072
BatchNorm2d-125 [-1, 256, 26, 26] 512
LeakyReLU-126 [-1, 256, 26, 26] 0
Conv2d-127 [-1, 512, 26, 26] 1,179,648
BatchNorm2d-128 [-1, 512, 26, 26] 1,024
LeakyReLU-129 [-1, 512, 26, 26] 0
Conv2d-130 [-1, 1024, 13, 13] 4,718,592
BatchNorm2d-131 [-1, 1024, 13, 13] 2,048
LeakyReLU-132 [-1, 1024, 13, 13] 0
Conv2d-133 [-1, 512, 13, 13] 524,288
BatchNorm2d-134 [-1, 512, 13, 13] 1,024
LeakyReLU-135 [-1, 512, 13, 13] 0
Conv2d-136 [-1, 1024, 13, 13] 4,718,592
BatchNorm2d-137 [-1, 1024, 13, 13] 2,048
LeakyReLU-138 [-1, 1024, 13, 13] 0
Conv2d-139 [-1, 512, 13, 13] 524,288
BatchNorm2d-140 [-1, 512, 13, 13] 1,024
LeakyReLU-141 [-1, 512, 13, 13] 0
Conv2d-142 [-1, 1024, 13, 13] 4,718,592
BatchNorm2d-143 [-1, 1024, 13, 13] 2,048
LeakyReLU-144 [-1, 1024, 13, 13] 0
Conv2d-145 [-1, 512, 13, 13] 524,288
BatchNorm2d-146 [-1, 512, 13, 13] 1,024
LeakyReLU-147 [-1, 512, 13, 13] 0
Conv2d-148 [-1, 1024, 13, 13] 4,718,592
BatchNorm2d-149 [-1, 1024, 13, 13] 2,048
LeakyReLU-150 [-1, 1024, 13, 13] 0
Conv2d-151 [-1, 512, 13, 13] 524,288
BatchNorm2d-152 [-1, 512, 13, 13] 1,024
LeakyReLU-153 [-1, 512, 13, 13] 0
Conv2d-154 [-1, 1024, 13, 13] 4,718,592
BatchNorm2d-155 [-1, 1024, 13, 13] 2,048
LeakyReLU-156 [-1, 1024, 13, 13] 0
Conv2d-157 [-1, 512, 13, 13] 524,288
BatchNorm2d-158 [-1, 512, 13, 13] 1,024
LeakyReLU-159 [-1, 512, 13, 13] 0
Conv2d-160 [-1, 1024, 13, 13] 4,718,592
BatchNorm2d-161 [-1, 1024, 13, 13] 2,048
LeakyReLU-162 [-1, 1024, 13, 13] 0
Conv2d-163 [-1, 512, 13, 13] 524,288
BatchNorm2d-164 [-1, 512, 13, 13] 1,024
LeakyReLU-165 [-1, 512, 13, 13] 0
Conv2d-166 [-1, 1024, 13, 13] 4,718,592
BatchNorm2d-167 [-1, 1024, 13, 13] 2,048
LeakyReLU-168 [-1, 1024, 13, 13] 0
Conv2d-169 [-1, 512, 13, 13] 524,288
BatchNorm2d-170 [-1, 512, 13, 13] 1,024
LeakyReLU-171 [-1, 512, 13, 13] 0
Conv2d-172 [-1, 1024, 13, 13] 4,718,592
BatchNorm2d-173 [-1, 1024, 13, 13] 2,048
LeakyReLU-174 [-1, 1024, 13, 13] 0
Conv2d-175 [-1, 255, 13, 13] 261,375
Conv2d-176 [-1, 256, 13, 13] 131,072
BatchNorm2d-177 [-1, 256, 13, 13] 512
LeakyReLU-178 [-1, 256, 13, 13] 0
Upsample-179 [-1, 256, 26, 26] 0
Conv2d-180 [-1, 256, 26, 26] 196,608
BatchNorm2d-181 [-1, 256, 26, 26] 512
LeakyReLU-182 [-1, 256, 26, 26] 0
Conv2d-183 [-1, 512, 26, 26] 1,179,648
BatchNorm2d-184 [-1, 512, 26, 26] 1,024
LeakyReLU-185 [-1, 512, 26, 26] 0
Conv2d-186 [-1, 256, 26, 26] 131,072
BatchNorm2d-187 [-1, 256, 26, 26] 512
LeakyReLU-188 [-1, 256, 26, 26] 0
Conv2d-189 [-1, 512, 26, 26] 1,179,648
BatchNorm2d-190 [-1, 512, 26, 26] 1,024
LeakyReLU-191 [-1, 512, 26, 26] 0
Conv2d-192 [-1, 256, 26, 26] 131,072
BatchNorm2d-193 [-1, 256, 26, 26] 512
LeakyReLU-194 [-1, 256, 26, 26] 0
Conv2d-195 [-1, 512, 26, 26] 1,179,648
BatchNorm2d-196 [-1, 512, 26, 26] 1,024
LeakyReLU-197 [-1, 512, 26, 26] 0
Conv2d-198 [-1, 255, 26, 26] 130,815
Conv2d-199 [-1, 128, 26, 26] 32,768
BatchNorm2d-200 [-1, 128, 26, 26] 256
LeakyReLU-201 [-1, 128, 26, 26] 0
Upsample-202 [-1, 128, 52, 52] 0
Conv2d-203 [-1, 128, 52, 52] 49,152
BatchNorm2d-204 [-1, 128, 52, 52] 256
LeakyReLU-205 [-1, 128, 52, 52] 0
Conv2d-206 [-1, 256, 52, 52] 294,912
BatchNorm2d-207 [-1, 256, 52, 52] 512
LeakyReLU-208 [-1, 256, 52, 52] 0
Conv2d-209 [-1, 128, 52, 52] 32,768
BatchNorm2d-210 [-1, 128, 52, 52] 256
LeakyReLU-211 [-1, 128, 52, 52] 0
Conv2d-212 [-1, 256, 52, 52] 294,912
BatchNorm2d-213 [-1, 256, 52, 52] 512
LeakyReLU-214 [-1, 256, 52, 52] 0
Conv2d-215 [-1, 128, 52, 52] 32,768
BatchNorm2d-216 [-1, 128, 52, 52] 256
LeakyReLU-217 [-1, 128, 52, 52] 0
Conv2d-218 [-1, 256, 52, 52] 294,912
BatchNorm2d-219 [-1, 256, 52, 52] 512
LeakyReLU-220 [-1, 256, 52, 52] 0
Conv2d-221 [-1, 255, 52, 52] 65,535
================================================================
Total params: 61,949,149
Trainable params: 61,949,149
Non-trainable params: 0
----------------------------------------------------------------