./build/x86_64/bin/trt_ssd image_filename model_prototxt model_weights mode Building and running a GPU inference engine for ./data/deploy_gie.prototxt, N=1... loaded image ./data/1.jpg (600 x 300) 2880000 bytes [cuda] cudaAllocMapped 2880000 bytes, CPU 0x200c00000 GPU 0x200c00000 Load in image ./data/1.jpg width: 600 height: 300 depth: 24 (bpp) NormalizeLayer start to getOutputDimensions! NormalizeLayer nbInputDims is 1 NormalizeLayer inputs[0].d[0,1,2] is 512,38,75 PermuteLayer inputs[0].d[0,1,2] is 16,38,75 FlattenLayer inputs[0].d[0,1,2] is 38,75,16 FlattenLayer outputs[0].d[0,1,2] is 45600,1,1 PermuteLayer inputs[0].d[0,1,2] is 8,38,75 FlattenLayer inputs[0].d[0,1,2] is 38,75,8 FlattenLayer outputs[0].d[0,1,2] is 22800,1,1 PermuteLayer inputs[0].d[0,1,2] is 24,19,38 FlattenLayer inputs[0].d[0,1,2] is 19,38,24 FlattenLayer outputs[0].d[0,1,2] is 17328,1,1 PermuteLayer inputs[0].d[0,1,2] is 12,19,38 FlattenLayer inputs[0].d[0,1,2] is 19,38,12 FlattenLayer outputs[0].d[0,1,2] is 8664,1,1 PermuteLayer inputs[0].d[0,1,2] is 24,10,19 FlattenLayer inputs[0].d[0,1,2] is 10,19,24 FlattenLayer outputs[0].d[0,1,2] is 4560,1,1 PermuteLayer inputs[0].d[0,1,2] is 12,10,19 FlattenLayer inputs[0].d[0,1,2] is 10,19,12 FlattenLayer outputs[0].d[0,1,2] is 2280,1,1 PermuteLayer inputs[0].d[0,1,2] is 24,5,10 FlattenLayer inputs[0].d[0,1,2] is 5,10,24 FlattenLayer outputs[0].d[0,1,2] is 1200,1,1 PermuteLayer inputs[0].d[0,1,2] is 12,5,10 FlattenLayer inputs[0].d[0,1,2] is 5,10,12 FlattenLayer outputs[0].d[0,1,2] is 600,1,1 PermuteLayer inputs[0].d[0,1,2] is 16,3,8 FlattenLayer inputs[0].d[0,1,2] is 3,8,16 FlattenLayer outputs[0].d[0,1,2] is 384,1,1 PermuteLayer inputs[0].d[0,1,2] is 8,3,8 FlattenLayer inputs[0].d[0,1,2] is 3,8,8 FlattenLayer outputs[0].d[0,1,2] is 192,1,1 PermuteLayer inputs[0].d[0,1,2] is 16,1,6 FlattenLayer inputs[0].d[0,1,2] is 1,6,16 FlattenLayer outputs[0].d[0,1,2] is 96,1,1 PermuteLayer inputs[0].d[0,1,2] is 8,1,6 FlattenLayer inputs[0].d[0,1,2] is 1,6,8 FlattenLayer outputs[0].d[0,1,2] is 48,1,1 ReshapeLayer inputs[0].d[0,1,2] is 34584,1,1 ReshapeLayer outputs[0].d[0,1,2] is 17292,2,1 SoftmaxOnHLayer inputs[0].d[0,1,2] is 17292,2,1 FlattenLayer inputs[0].d[0,1,2] is 17292,2,1 FlattenLayer outputs[0].d[0,1,2] is 34584,1,1 NormalizeLayer start to configure! inputs[0].d[0,1,2] is 512,38,75 PermuteLayer inputs[0].d[0,1,2] is 16,38,75 PermuteLayer outputs[0].d[0,1,2] is 38,75,16 PermuteLayer inputs[0].d[0,1,2] is 8,38,75 PermuteLayer outputs[0].d[0,1,2] is 38,75,8 PermuteLayer inputs[0].d[0,1,2] is 24,19,38 PermuteLayer outputs[0].d[0,1,2] is 19,38,24 PermuteLayer inputs[0].d[0,1,2] is 12,19,38 PermuteLayer outputs[0].d[0,1,2] is 19,38,12 PermuteLayer inputs[0].d[0,1,2] is 24,10,19 PermuteLayer outputs[0].d[0,1,2] is 10,19,24 PermuteLayer inputs[0].d[0,1,2] is 12,10,19 PermuteLayer outputs[0].d[0,1,2] is 10,19,12 PermuteLayer inputs[0].d[0,1,2] is 24,5,10 PermuteLayer outputs[0].d[0,1,2] is 5,10,24 PermuteLayer inputs[0].d[0,1,2] is 12,5,10 PermuteLayer outputs[0].d[0,1,2] is 5,10,12 PermuteLayer inputs[0].d[0,1,2] is 16,3,8 PermuteLayer outputs[0].d[0,1,2] is 3,8,16 PermuteLayer inputs[0].d[0,1,2] is 8,3,8 PermuteLayer outputs[0].d[0,1,2] is 3,8,8 PermuteLayer inputs[0].d[0,1,2] is 16,1,6 PermuteLayer outputs[0].d[0,1,2] is 1,6,16 PermuteLayer inputs[0].d[0,1,2] is 8,1,6 PermuteLayer outputs[0].d[0,1,2] is 1,6,8 SoftmaxOnHLayer inputs[0].d[0,1,2] is 17292,2,1 SoftmaxOnHLayer outputs[0].d[0,1,2] is 17292,2,1 NormalizeLayer start to initialize! mOldStepsCPU[0] is 45600, mNewStepsCPU[0] is 45600 mOldStepsCPU[1] is 2850, mNewStepsCPU[1] is 1200 mOldStepsCPU[2] is 75, mNewStepsCPU[2] is 16 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 mOldStepsCPU[0] is 22800, mNewStepsCPU[0] is 22800 mOldStepsCPU[1] is 2850, mNewStepsCPU[1] is 600 mOldStepsCPU[2] is 75, mNewStepsCPU[2] is 8 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 mOldStepsCPU[0] is 17328, mNewStepsCPU[0] is 17328 mOldStepsCPU[1] is 722, mNewStepsCPU[1] is 912 mOldStepsCPU[2] is 38, mNewStepsCPU[2] is 24 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 mOldStepsCPU[0] is 8664, mNewStepsCPU[0] is 8664 mOldStepsCPU[1] is 722, mNewStepsCPU[1] is 456 mOldStepsCPU[2] is 38, mNewStepsCPU[2] is 12 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 mOldStepsCPU[0] is 4560, mNewStepsCPU[0] is 4560 mOldStepsCPU[1] is 190, mNewStepsCPU[1] is 456 mOldStepsCPU[2] is 19, mNewStepsCPU[2] is 24 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 mOldStepsCPU[0] is 2280, mNewStepsCPU[0] is 2280 mOldStepsCPU[1] is 190, mNewStepsCPU[1] is 228 mOldStepsCPU[2] is 19, mNewStepsCPU[2] is 12 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 mOldStepsCPU[0] is 1200, mNewStepsCPU[0] is 1200 mOldStepsCPU[1] is 50, mNewStepsCPU[1] is 240 mOldStepsCPU[2] is 10, mNewStepsCPU[2] is 24 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 mOldStepsCPU[0] is 600, mNewStepsCPU[0] is 600 mOldStepsCPU[1] is 50, mNewStepsCPU[1] is 120 mOldStepsCPU[2] is 10, mNewStepsCPU[2] is 12 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 mOldStepsCPU[0] is 384, mNewStepsCPU[0] is 384 mOldStepsCPU[1] is 24, mNewStepsCPU[1] is 128 mOldStepsCPU[2] is 8, mNewStepsCPU[2] is 16 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 mOldStepsCPU[0] is 192, mNewStepsCPU[0] is 192 mOldStepsCPU[1] is 24, mNewStepsCPU[1] is 64 mOldStepsCPU[2] is 8, mNewStepsCPU[2] is 8 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 mOldStepsCPU[0] is 96, mNewStepsCPU[0] is 96 mOldStepsCPU[1] is 6, mNewStepsCPU[1] is 96 mOldStepsCPU[2] is 6, mNewStepsCPU[2] is 16 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 mOldStepsCPU[0] is 48, mNewStepsCPU[0] is 48 mOldStepsCPU[1] is 6, mNewStepsCPU[1] is 48 mOldStepsCPU[2] is 6, mNewStepsCPU[2] is 8 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 SoftmaxOnHLayer start to initialize! NormalizeLayer start to getSerializationSize! NormalizeLayer start to serialize! NormalizeLayer start to getSerializationSize! NormalizeLayer start to getSerializationSize! NormalizeLayer start to serialize! NormalizeLayer start to terminate! SoftmaxOnHLayer start to terminate! NormalizeLayer size is 2060 PermuteLayer constructed from Serialization with size 28 PermuteLayer constructed from Serialization with size 28 PermuteLayer constructed from Serialization with size 28 PermuteLayer constructed from Serialization with size 28 PermuteLayer constructed from Serialization with size 28 PermuteLayer constructed from Serialization with size 28 PermuteLayer constructed from Serialization with size 28 PermuteLayer constructed from Serialization with size 28 PermuteLayer constructed from Serialization with size 28 PermuteLayer constructed from Serialization with size 28 PermuteLayer constructed from Serialization with size 28 PermuteLayer constructed from Serialization with size 28 NormalizeLayer start to initialize! mOldStepsCPU[0] is 45600, mNewStepsCPU[0] is 45600 mOldStepsCPU[1] is 2850, mNewStepsCPU[1] is 1200 mOldStepsCPU[2] is 75, mNewStepsCPU[2] is 16 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 mOldStepsCPU[0] is 22800, mNewStepsCPU[0] is 22800 mOldStepsCPU[1] is 2850, mNewStepsCPU[1] is 600 mOldStepsCPU[2] is 75, mNewStepsCPU[2] is 8 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 mOldStepsCPU[0] is 17328, mNewStepsCPU[0] is 17328 mOldStepsCPU[1] is 722, mNewStepsCPU[1] is 912 mOldStepsCPU[2] is 38, mNewStepsCPU[2] is 24 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 mOldStepsCPU[0] is 8664, mNewStepsCPU[0] is 8664 mOldStepsCPU[1] is 722, mNewStepsCPU[1] is 456 mOldStepsCPU[2] is 38, mNewStepsCPU[2] is 12 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 mOldStepsCPU[0] is 4560, mNewStepsCPU[0] is 4560 mOldStepsCPU[1] is 190, mNewStepsCPU[1] is 456 mOldStepsCPU[2] is 19, mNewStepsCPU[2] is 24 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 mOldStepsCPU[0] is 2280, mNewStepsCPU[0] is 2280 mOldStepsCPU[1] is 190, mNewStepsCPU[1] is 228 mOldStepsCPU[2] is 19, mNewStepsCPU[2] is 12 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 mOldStepsCPU[0] is 1200, mNewStepsCPU[0] is 1200 mOldStepsCPU[1] is 50, mNewStepsCPU[1] is 240 mOldStepsCPU[2] is 10, mNewStepsCPU[2] is 24 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 mOldStepsCPU[0] is 600, mNewStepsCPU[0] is 600 mOldStepsCPU[1] is 50, mNewStepsCPU[1] is 120 mOldStepsCPU[2] is 10, mNewStepsCPU[2] is 12 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 mOldStepsCPU[0] is 384, mNewStepsCPU[0] is 384 mOldStepsCPU[1] is 24, mNewStepsCPU[1] is 128 mOldStepsCPU[2] is 8, mNewStepsCPU[2] is 16 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 mOldStepsCPU[0] is 192, mNewStepsCPU[0] is 192 mOldStepsCPU[1] is 24, mNewStepsCPU[1] is 64 mOldStepsCPU[2] is 8, mNewStepsCPU[2] is 8 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 mOldStepsCPU[0] is 96, mNewStepsCPU[0] is 96 mOldStepsCPU[1] is 6, mNewStepsCPU[1] is 96 mOldStepsCPU[2] is 6, mNewStepsCPU[2] is 16 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 mOldStepsCPU[0] is 48, mNewStepsCPU[0] is 48 mOldStepsCPU[1] is 6, mNewStepsCPU[1] is 48 mOldStepsCPU[2] is 6, mNewStepsCPU[2] is 8 mOldStepsCPU[3] is 1, mNewStepsCPU[3] is 1 SoftmaxOnHLayer start to initialize! Bindings after deserializing: Binding 0 (data): Input. Binding 1 (mbox_loc): Output. Binding 2 (mbox_conf_flatten): Output. Allocate memory: data with 540000 float numbers Allocate memory: mbox_loc with 69168 float numbers Allocate memory: mbox_conf_flatten with 34584 float numbers Allocate memory: prior_data with 138336 float numbers Allocate memory: bbox_data_unified_memory with 69168 float numbers Allocate memory: conf_permute_unified_memory with 34584 float numbers start main loop SoftmaxOnHLayer start to enqueue! Forward SoftmaxOnHLayer layer to end! main graph flow complete total_num_priors = 17292 Wait GPU work is done DetectionOutputLayer CPU part consumes 0.346689ms detection output final layer complete 0--1, 0.996268, 0.179394, 0.267203, 0.284639, 0.846807 0--1, 0.99345, 0.781464, 0.256652, 0.900316, 0.852235 0--1, 0.993077, 0.0339053, 0.271945, 0.15262, 0.845425 0--1, 0.988621, 0.447997, 0.263895, 0.548257, 0.842531 0--1, 0.948939, 0.306625, 0.258145, 0.413164, 0.846016 0--1, 0.948452, 0.575053, 0.257146, 0.678382, 0.849395 0--1, 0.687262, 0.936254, 0.00462355, 1.01454, 0.17509 conv1_1_v input reformatter 0 0.290ms conv1_1_v 0.096ms conv1_1_h + relu1_1 0.164ms conv1_2_v 0.207ms conv1_2_h + relu1_2 0.233ms pool1 0.146ms conv2_1_v 0.104ms conv2_1_h + relu2_1 0.164ms conv2_2_v 0.159ms conv2_2_h + relu2_2 0.165ms pool2 0.077ms conv3_1_v 0.057ms conv3_1_h + relu3_1 0.096ms conv3_2_v 0.152ms conv3_2_h + relu3_2 0.151ms conv3_3_v 0.205ms conv3_3_h + relu3_3 0.194ms pool3 0.043ms conv4_1_v 0.074ms conv4_1_h + relu4_1 0.120ms conv4_2_v 0.126ms conv4_2_h + relu4_2 0.120ms conv4_3_v 0.156ms conv4_3_h + relu4_3 0.150ms pool4 0.027ms conv5_1_v 0.074ms conv5_1_h + relu5_1 0.071ms conv5_2_v 0.068ms conv5_2_h + relu5_2 0.067ms conv5_3_v 0.067ms conv5_3_h + relu5_3 0.066ms pool5 0.018ms fc6_v input reformatter 0 0.022ms fc6_v 0.463ms fc6_h + relu6 0.990ms fc7_v input reformatter 0 0.030ms fc7_v 0.074ms fc7_h + relu7 0.066ms fc7_h + relu7 output reformatter 0 0.039ms conv6_1 + conv6_1_relu input reformatter 0.032ms conv6_1 + conv6_1_relu 0.051ms conv6_2 + conv6_2_relu 0.086ms conv6_2 + conv6_2_relu output reformatte 0.010ms conv7_1 + conv7_1_relu input reformatter 0.007ms conv7_1 + conv7_1_relu 0.015ms conv7_2 + conv7_2_relu 0.049ms conv7_2 + conv7_2_relu output reformatte 0.006ms conv8_1 + conv8_1_relu input reformatter 0.005ms conv8_1 + conv8_1_relu 0.012ms conv8_2 + conv8_2_relu 0.019ms conv8_2 + conv8_2_relu output reformatte 0.006ms conv9_1 + conv9_1_relu input reformatter 0.005ms conv9_1 + conv9_1_relu 0.011ms conv9_2 + conv9_2_relu 0.015ms conv9_2 + conv9_2_relu output reformatte 0.005ms conv4_3_norm input reformatter 0 0.069ms conv4_3_norm 0.323ms conv4_3_norm_mbox_loc 0.173ms conv4_3_norm_mbox_loc_perm 0.020ms conv4_3_norm_mbox_loc_flat 0.005ms conv4_3_norm_mbox_conf 0.172ms conv4_3_norm_mbox_conf_perm 0.011ms conv4_3_norm_mbox_conf_flat 0.004ms fc7_mbox_loc 0.194ms fc7_mbox_loc_perm 0.009ms fc7_mbox_loc_flat 0.004ms fc7_mbox_conf 0.194ms fc7_mbox_conf_perm 0.007ms fc7_mbox_conf_flat 0.004ms conv6_2_mbox_loc 0.103ms conv6_2_mbox_loc_perm 0.007ms conv6_2_mbox_loc_flat 0.004ms conv6_2_mbox_conf 0.089ms conv6_2_mbox_conf_perm 0.007ms conv6_2_mbox_conf_flat 0.004ms conv7_2_mbox_loc 0.054ms conv7_2_mbox_loc_perm 0.007ms conv7_2_mbox_loc_flat 0.004ms conv7_2_mbox_conf 0.051ms conv7_2_mbox_conf_perm 0.006ms conv7_2_mbox_conf_flat 0.004ms conv8_2_mbox_loc 0.052ms conv8_2_mbox_loc_perm 0.006ms conv8_2_mbox_loc_flat 0.004ms conv8_2_mbox_conf 0.052ms conv8_2_mbox_conf_perm 0.007ms conv8_2_mbox_conf_flat 0.003ms conv9_2_mbox_loc 0.053ms conv9_2_mbox_loc_perm 0.006ms conv9_2_mbox_loc_flat 0.004ms conv9_2_mbox_conf 0.054ms conv9_2_mbox_conf_perm 0.006ms conv9_2_mbox_conf_flat 0.004ms conv4_3_norm_mbox_loc_flat copy 0.176ms fc7_mbox_loc_flat copy 0.167ms conv6_2_mbox_loc_flat copy 0.004ms conv7_2_mbox_loc_flat copy 0.004ms conv8_2_mbox_loc_flat copy 0.004ms conv9_2_mbox_loc_flat copy 0.004ms conv4_3_norm_mbox_conf_flat copy 0.004ms fc7_mbox_conf_flat copy 0.004ms conv6_2_mbox_conf_flat copy 0.004ms conv7_2_mbox_conf_flat copy 0.004ms conv8_2_mbox_conf_flat copy 0.004ms conv9_2_mbox_conf_flat copy 0.004ms mbox_conf_reshape 0.004ms mbox_conf_softmax 0.025ms mbox_conf_flatten 0.352ms Time over all layers: 8.471 NormalizeLayer start to terminate! SoftmaxOnHLayer start to terminate! Done.