Conversion from caffemodel to TensorRT

Hello everyone,
i am trying to convert some caffemodels to tensorRt, my first question is if a model contains all the supported operations is it able to convert?
In particular I am trying to convert this model:
res10_300x300_ssd_iter_140000_fp16.caffemodel, but it returns me this error "log (): mbox_loc: all concat input tensors must have the same dimensions except on the concatenation axis (0), but dimensions mismatched at input 1 at index 1. Input 0 shape: [38 , 38,16], Input 1 shape: [19,19,24] "is there anyone who has any idea about it?

Hi,

Could you try your model with trtexec and share the corresponding log with us?

/usr/src/tensorrt/bin/trtexec --deploy=[your_model.prototxt] --output=[output_layer_name] --verbose

Thanks.

&&&& RUNNING TensorRT.trtexec # /usr/src/tensorrt/bin/trtexec --deploy=/home/dlinano/deepstream_sdk_v4.0.2_jetson/sources/apps/redaction_with_deepstream/fd_lpd_model/face_deploy.prototxt --output=detection_out --verbose
[02/17/2020-02:19:01] [I] === Model Options ===
[02/17/2020-02:19:01] [I] Format: Caffe
[02/17/2020-02:19:01] [I] Model:
[02/17/2020-02:19:01] [I] Prototxt: /home/dlinano/deepstream_sdk_v4.0.2_jetson/sources/apps/redaction_with_deepstream/fd_lpd_model/face_deploy.prototxtOutput: detection_out
[02/17/2020-02:19:01] [I] === Build Options ===
[02/17/2020-02:19:01] [I] Max batch: 1
[02/17/2020-02:19:01] [I] Workspace: 16 MB
[02/17/2020-02:19:01] [I] minTiming: 1
[02/17/2020-02:19:01] [I] avgTiming: 8
[02/17/2020-02:19:01] [I] Precision: FP32
[02/17/2020-02:19:01] [I] Calibration:
[02/17/2020-02:19:01] [I] Safe mode: Disabled
[02/17/2020-02:19:01] [I] Save engine:
[02/17/2020-02:19:01] [I] Load engine:
[02/17/2020-02:19:01] [I] Inputs format: fp32:CHW
[02/17/2020-02:19:01] [I] Outputs format: fp32:CHW
[02/17/2020-02:19:01] [I] Input build shapes: model
[02/17/2020-02:19:01] [I] === System Options ===
[02/17/2020-02:19:01] [I] Device: 0
[02/17/2020-02:19:01] [I] DLACore:
[02/17/2020-02:19:01] [I] Plugins:
[02/17/2020-02:19:01] [I] === Inference Options ===
[02/17/2020-02:19:01] [I] Batch: 1
[02/17/2020-02:19:01] [I] Iterations: 10 (200 ms warm up)
[02/17/2020-02:19:01] [I] Duration: 10s
[02/17/2020-02:19:01] [I] Sleep time: 0ms
[02/17/2020-02:19:01] [I] Streams: 1
[02/17/2020-02:19:01] [I] Spin-wait: Disabled
[02/17/2020-02:19:01] [I] Multithreading: Enabled
[02/17/2020-02:19:01] [I] CUDA Graph: Disabled
[02/17/2020-02:19:01] [I] Skip inference: Disabled
[02/17/2020-02:19:01] [I] Input inference shapes: model
[02/17/2020-02:19:01] [I] === Reporting Options ===
[02/17/2020-02:19:01] [I] Verbose: Enabled
[02/17/2020-02:19:01] [I] Averages: 10 inferences
[02/17/2020-02:19:01] [I] Percentile: 99
[02/17/2020-02:19:01] [I] Dump output: Disabled
[02/17/2020-02:19:01] [I] Profile: Disabled
[02/17/2020-02:19:01] [I] Export timing to JSON file:
[02/17/2020-02:19:01] [I] Export profile to JSON file:
[02/17/2020-02:19:01] [I]
[02/17/2020-02:19:01] [V] [TRT] Plugin Creator registration succeeded - GridAnchor_TRT
[02/17/2020-02:19:01] [V] [TRT] Plugin Creator registration succeeded - GridAnchorRect_TRT
[02/17/2020-02:19:01] [V] [TRT] Plugin Creator registration succeeded - NMS_TRT
[02/17/2020-02:19:01] [V] [TRT] Plugin Creator registration succeeded - Reorg_TRT
[02/17/2020-02:19:01] [V] [TRT] Plugin Creator registration succeeded - Region_TRT
[02/17/2020-02:19:01] [V] [TRT] Plugin Creator registration succeeded - Clip_TRT
[02/17/2020-02:19:01] [V] [TRT] Plugin Creator registration succeeded - LReLU_TRT
[02/17/2020-02:19:01] [V] [TRT] Plugin Creator registration succeeded - PriorBox_TRT
[02/17/2020-02:19:01] [V] [TRT] Plugin Creator registration succeeded - Normalize_TRT
[02/17/2020-02:19:01] [V] [TRT] Plugin Creator registration succeeded - RPROI_TRT
[02/17/2020-02:19:01] [V] [TRT] Plugin Creator registration succeeded - BatchedNMS_TRT
[02/17/2020-02:19:01] [V] [TRT] Plugin Creator registration succeeded - FlattenConcat_TRT
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
Warning: Flatten layer ignored. TensorRT implicitly flattens input to FullyConnected layers, but in other circumstances this will result in undefined behavior.
[02/17/2020-02:19:02] [E] [TRT] mbox_loc: all concat input tensors must have the same dimensions except on the concatenation axis (0), but dimensions mismatched at input 1 at index 1. Input 0 shape: [38,38,16], Input 1 shape: [19,19,24]
[02/17/2020-02:19:02] [E] [TRT] mbox_conf: all concat input tensors must have the same dimensions except on the concatenation axis (0), but dimensions mismatched at input 1 at index 1. Input 0 shape: [38,38,8], Input 1 shape: [19,19,12]
Caffe Parser: Invalid axis in softmax layer - TensorRT expects NCHW input. Negative axis is not supported in TensorRT, please use positive axis indexing
error parsing layer type Softmax index 109
[02/17/2020-02:19:02] [E] Failed to parse caffe model or prototxt, tensors blob not found
[02/17/2020-02:19:02] [E] Parsing model failed
[02/17/2020-02:19:02] [E] Engine could not be created
&&&& FAILED TensorRT.trtexec # /usr/src/tensorrt/bin/trtexec --deploy=/home/dlinano/deepstream_sdk_v4.0.2_jetson/sources/apps/redaction_with_deepstream/fd_lpd_model/face_deploy.prototxt --output=detection_out --verbose

Hi,

Sorry for the late update.

Our Caffe parser doesn’t support Flatten layer.
However, you can use Reshape layer to do the identical transform instead.

reshape_param {
  shape {
    dim: 0
    dim: -1
    dim: 1
    dim: 1
  }
}

There is another issue in your model, please add one more output in the detection_out layer to match the DetectionOutput plugin requirement.

top: "keep_count"

A similar change can also be found in our SSD sample:

/usr/src/tensorrt/samples/sampleSSD/

Expecting you are using the model shared in the below GitHub, we also make an workable deploy.prototxt for you:
https://github.com/spmallick/learnopencv/tree/master/FaceDetectionComparison/models

input: "data"
input_shape {
  dim: 1
  dim: 3
  dim: 300
  dim: 300
}

layer {
  name: "data_bn"
  type: "BatchNorm"
  bottom: "data"
  top: "data_bn"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "data_scale"
  type: "Scale"
  bottom: "data_bn"
  top: "data_bn"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "conv1_h"
  type: "Convolution"
  bottom: "data_bn"
  top: "conv1_h"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 32
    pad: 3
    kernel_size: 7
    stride: 2
    weight_filler {
      type: "msra"
      variance_norm: FAN_OUT
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "conv1_bn_h"
  type: "BatchNorm"
  bottom: "conv1_h"
  top: "conv1_h"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "conv1_scale_h"
  type: "Scale"
  bottom: "conv1_h"
  top: "conv1_h"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "conv1_relu"
  type: "ReLU"
  bottom: "conv1_h"
  top: "conv1_h"
}
layer {
  name: "conv1_pool"
  type: "Pooling"
  bottom: "conv1_h"
  top: "conv1_pool"
  pooling_param {
    kernel_size: 3
    stride: 2
  }
}
layer {
  name: "layer_64_1_conv1_h"
  type: "Convolution"
  bottom: "conv1_pool"
  top: "layer_64_1_conv1_h"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 32
    bias_term: false
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_64_1_bn2_h"
  type: "BatchNorm"
  bottom: "layer_64_1_conv1_h"
  top: "layer_64_1_conv1_h"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_64_1_scale2_h"
  type: "Scale"
  bottom: "layer_64_1_conv1_h"
  top: "layer_64_1_conv1_h"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_64_1_relu2"
  type: "ReLU"
  bottom: "layer_64_1_conv1_h"
  top: "layer_64_1_conv1_h"
}
layer {
  name: "layer_64_1_conv2_h"
  type: "Convolution"
  bottom: "layer_64_1_conv1_h"
  top: "layer_64_1_conv2_h"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 32
    bias_term: false
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_64_1_sum"
  type: "Eltwise"
  bottom: "layer_64_1_conv2_h"
  bottom: "conv1_pool"
  top: "layer_64_1_sum"
}
layer {
  name: "layer_128_1_bn1_h"
  type: "BatchNorm"
  bottom: "layer_64_1_sum"
  top: "layer_128_1_bn1_h"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_128_1_scale1_h"
  type: "Scale"
  bottom: "layer_128_1_bn1_h"
  top: "layer_128_1_bn1_h"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_128_1_relu1"
  type: "ReLU"
  bottom: "layer_128_1_bn1_h"
  top: "layer_128_1_bn1_h"
}
layer {
  name: "layer_128_1_conv1_h"
  type: "Convolution"
  bottom: "layer_128_1_bn1_h"
  top: "layer_128_1_conv1_h"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 128
    bias_term: false
    pad: 1
    kernel_size: 3
    stride: 2
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_128_1_bn2"
  type: "BatchNorm"
  bottom: "layer_128_1_conv1_h"
  top: "layer_128_1_conv1_h"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_128_1_scale2"
  type: "Scale"
  bottom: "layer_128_1_conv1_h"
  top: "layer_128_1_conv1_h"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_128_1_relu2"
  type: "ReLU"
  bottom: "layer_128_1_conv1_h"
  top: "layer_128_1_conv1_h"
}
layer {
  name: "layer_128_1_conv2"
  type: "Convolution"
  bottom: "layer_128_1_conv1_h"
  top: "layer_128_1_conv2"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 128
    bias_term: false
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_128_1_conv_expand_h"
  type: "Convolution"
  bottom: "layer_128_1_bn1_h"
  top: "layer_128_1_conv_expand_h"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 128
    bias_term: false
    pad: 0
    kernel_size: 1
    stride: 2
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_128_1_sum"
  type: "Eltwise"
  bottom: "layer_128_1_conv2"
  bottom: "layer_128_1_conv_expand_h"
  top: "layer_128_1_sum"
}
layer {
  name: "layer_256_1_bn1"
  type: "BatchNorm"
  bottom: "layer_128_1_sum"
  top: "layer_256_1_bn1"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_256_1_scale1"
  type: "Scale"
  bottom: "layer_256_1_bn1"
  top: "layer_256_1_bn1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_256_1_relu1"
  type: "ReLU"
  bottom: "layer_256_1_bn1"
  top: "layer_256_1_bn1"
}
layer {
  name: "layer_256_1_conv1"
  type: "Convolution"
  bottom: "layer_256_1_bn1"
  top: "layer_256_1_conv1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 256
    bias_term: false
    pad: 1
    kernel_size: 3
    stride: 2
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_256_1_bn2"
  type: "BatchNorm"
  bottom: "layer_256_1_conv1"
  top: "layer_256_1_conv1"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_256_1_scale2"
  type: "Scale"
  bottom: "layer_256_1_conv1"
  top: "layer_256_1_conv1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_256_1_relu2"
  type: "ReLU"
  bottom: "layer_256_1_conv1"
  top: "layer_256_1_conv1"
}
layer {
  name: "layer_256_1_conv2"
  type: "Convolution"
  bottom: "layer_256_1_conv1"
  top: "layer_256_1_conv2"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 256
    bias_term: false
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_256_1_conv_expand"
  type: "Convolution"
  bottom: "layer_256_1_bn1"
  top: "layer_256_1_conv_expand"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 256
    bias_term: false
    pad: 0
    kernel_size: 1
    stride: 2
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_256_1_sum"
  type: "Eltwise"
  bottom: "layer_256_1_conv2"
  bottom: "layer_256_1_conv_expand"
  top: "layer_256_1_sum"
}
layer {
  name: "layer_512_1_bn1"
  type: "BatchNorm"
  bottom: "layer_256_1_sum"
  top: "layer_512_1_bn1"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_512_1_scale1"
  type: "Scale"
  bottom: "layer_512_1_bn1"
  top: "layer_512_1_bn1"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_512_1_relu1"
  type: "ReLU"
  bottom: "layer_512_1_bn1"
  top: "layer_512_1_bn1"
}
layer {
  name: "layer_512_1_conv1_h"
  type: "Convolution"
  bottom: "layer_512_1_bn1"
  top: "layer_512_1_conv1_h"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 128
    bias_term: false
    pad: 1
    kernel_size: 3
    stride: 1 # 2
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_512_1_bn2_h"
  type: "BatchNorm"
  bottom: "layer_512_1_conv1_h"
  top: "layer_512_1_conv1_h"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "layer_512_1_scale2_h"
  type: "Scale"
  bottom: "layer_512_1_conv1_h"
  top: "layer_512_1_conv1_h"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "layer_512_1_relu2"
  type: "ReLU"
  bottom: "layer_512_1_conv1_h"
  top: "layer_512_1_conv1_h"
}
layer {
  name: "layer_512_1_conv2_h"
  type: "Convolution"
  bottom: "layer_512_1_conv1_h"
  top: "layer_512_1_conv2_h"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 256
    bias_term: false
    pad: 2 # 1
    kernel_size: 3
    stride: 1
    dilation: 2
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_512_1_conv_expand_h"
  type: "Convolution"
  bottom: "layer_512_1_bn1"
  top: "layer_512_1_conv_expand_h"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  convolution_param {
    num_output: 256
    bias_term: false
    pad: 0
    kernel_size: 1
    stride: 1 # 2
    weight_filler {
      type: "msra"
    }
    bias_filler {
      type: "constant"
      value: 0.0
    }
  }
}
layer {
  name: "layer_512_1_sum"
  type: "Eltwise"
  bottom: "layer_512_1_conv2_h"
  bottom: "layer_512_1_conv_expand_h"
  top: "layer_512_1_sum"
}
layer {
  name: "last_bn_h"
  type: "BatchNorm"
  bottom: "layer_512_1_sum"
  top: "layer_512_1_sum"
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
  param {
    lr_mult: 0.0
  }
}
layer {
  name: "last_scale_h"
  type: "Scale"
  bottom: "layer_512_1_sum"
  top: "layer_512_1_sum"
  param {
    lr_mult: 1.0
    decay_mult: 1.0
  }
  param {
    lr_mult: 2.0
    decay_mult: 1.0
  }
  scale_param {
    bias_term: true
  }
}
layer {
  name: "last_relu"
  type: "ReLU"
  bottom: "layer_512_1_sum"
  top: "fc7"
}

layer {
  name: "conv6_1_h"
  type: "Convolution"
  bottom: "fc7"
  top: "conv6_1_h"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 128
    pad: 0
    kernel_size: 1
    stride: 1
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv6_1_relu"
  type: "ReLU"
  bottom: "conv6_1_h"
  top: "conv6_1_h"
}
layer {
  name: "conv6_2_h"
  type: "Convolution"
  bottom: "conv6_1_h"
  top: "conv6_2_h"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 256
    pad: 1
    kernel_size: 3
    stride: 2
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv6_2_relu"
  type: "ReLU"
  bottom: "conv6_2_h"
  top: "conv6_2_h"
}
layer {
  name: "conv7_1_h"
  type: "Convolution"
  bottom: "conv6_2_h"
  top: "conv7_1_h"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 64
    pad: 0
    kernel_size: 1
    stride: 1
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv7_1_relu"
  type: "ReLU"
  bottom: "conv7_1_h"
  top: "conv7_1_h"
}
layer {
  name: "conv7_2_h"
  type: "Convolution"
  bottom: "conv7_1_h"
  top: "conv7_2_h"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 128
    pad: 1
    kernel_size: 3
    stride: 2
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv7_2_relu"
  type: "ReLU"
  bottom: "conv7_2_h"
  top: "conv7_2_h"
}
layer {
  name: "conv8_1_h"
  type: "Convolution"
  bottom: "conv7_2_h"
  top: "conv8_1_h"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 64
    pad: 0
    kernel_size: 1
    stride: 1
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv8_1_relu"
  type: "ReLU"
  bottom: "conv8_1_h"
  top: "conv8_1_h"
}
layer {
  name: "conv8_2_h"
  type: "Convolution"
  bottom: "conv8_1_h"
  top: "conv8_2_h"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 128
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv8_2_relu"
  type: "ReLU"
  bottom: "conv8_2_h"
  top: "conv8_2_h"
}
layer {
  name: "conv9_1_h"
  type: "Convolution"
  bottom: "conv8_2_h"
  top: "conv9_1_h"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 64
    pad: 0
    kernel_size: 1
    stride: 1
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv9_1_relu"
  type: "ReLU"
  bottom: "conv9_1_h"
  top: "conv9_1_h"
}
layer {
  name: "conv9_2_h"
  type: "Convolution"
  bottom: "conv9_1_h"
  top: "conv9_2_h"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 128
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv9_2_relu"
  type: "ReLU"
  bottom: "conv9_2_h"
  top: "conv9_2_h"
}
layer {
  name: "conv4_3_norm"
  type: "Normalize"
  bottom: "layer_256_1_bn1"
  top: "conv4_3_norm"
  norm_param {
    across_spatial: false
    scale_filler {
      type: "constant"
      value: 20
    }
    channel_shared: false
  }
}
layer {
  name: "conv4_3_norm_mbox_loc"
  type: "Convolution"
  bottom: "conv4_3_norm"
  top: "conv4_3_norm_mbox_loc"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 16
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv4_3_norm_mbox_loc_perm"
  type: "Permute"
  bottom: "conv4_3_norm_mbox_loc"
  top: "conv4_3_norm_mbox_loc_perm"
  permute_param {
    order: 0
    order: 2
    order: 3
    order: 1
  }
}
layer {
  name: "conv4_3_norm_mbox_loc_flat"
  type: "Reshape"
  bottom: "conv4_3_norm_mbox_loc_perm"
  top: "conv4_3_norm_mbox_loc_flat"
  reshape_param {
    shape {
      dim: 0
      dim: -1
      dim: 1
      dim: 1
    }
  }
}
layer {
  name: "conv4_3_norm_mbox_conf"
  type: "Convolution"
  bottom: "conv4_3_norm"
  top: "conv4_3_norm_mbox_conf"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 8 # 84
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv4_3_norm_mbox_conf_perm"
  type: "Permute"
  bottom: "conv4_3_norm_mbox_conf"
  top: "conv4_3_norm_mbox_conf_perm"
  permute_param {
    order: 0
    order: 2
    order: 3
    order: 1
  }
}
layer {
  name: "conv4_3_norm_mbox_conf_flat"
  type: "Reshape"
  bottom: "conv4_3_norm_mbox_conf_perm"
  top: "conv4_3_norm_mbox_conf_flat"
  reshape_param {
    shape {
      dim: 0
      dim: -1
      dim: 1
      dim: 1
    }
  }
}
layer {
  name: "conv4_3_norm_mbox_priorbox"
  type: "PriorBox"
  bottom: "conv4_3_norm"
  bottom: "data"
  top: "conv4_3_norm_mbox_priorbox"
  prior_box_param {
    min_size: 30.0
    max_size: 60.0
    aspect_ratio: 2
    flip: true
    clip: false
    variance: 0.1
    variance: 0.1
    variance: 0.2
    variance: 0.2
    step: 8
    offset: 0.5
  }
}
layer {
  name: "fc7_mbox_loc"
  type: "Convolution"
  bottom: "fc7"
  top: "fc7_mbox_loc"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 24
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "fc7_mbox_loc_perm"
  type: "Permute"
  bottom: "fc7_mbox_loc"
  top: "fc7_mbox_loc_perm"
  permute_param {
    order: 0
    order: 2
    order: 3
    order: 1
  }
}
layer {
  name: "fc7_mbox_loc_flat"
  type: "Reshape"
  bottom: "fc7_mbox_loc_perm"
  top: "fc7_mbox_loc_flat"
  reshape_param {
    shape {
      dim: 0
      dim: -1
      dim: 1
      dim: 1
    }
  }
}
layer {
  name: "fc7_mbox_conf"
  type: "Convolution"
  bottom: "fc7"
  top: "fc7_mbox_conf"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 12 # 126
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "fc7_mbox_conf_perm"
  type: "Permute"
  bottom: "fc7_mbox_conf"
  top: "fc7_mbox_conf_perm"
  permute_param {
    order: 0
    order: 2
    order: 3
    order: 1
  }
}
layer {
  name: "fc7_mbox_conf_flat"
  type: "Reshape"
  bottom: "fc7_mbox_conf_perm"
  top: "fc7_mbox_conf_flat"
  reshape_param {
    shape {
      dim: 0
      dim: -1
      dim: 1
      dim: 1
    }
  }
}
layer {
  name: "fc7_mbox_priorbox"
  type: "PriorBox"
  bottom: "fc7"
  bottom: "data"
  top: "fc7_mbox_priorbox"
  prior_box_param {
    min_size: 60.0
    max_size: 111.0
    aspect_ratio: 2
    aspect_ratio: 3
    flip: true
    clip: false
    variance: 0.1
    variance: 0.1
    variance: 0.2
    variance: 0.2
    step: 16
    offset: 0.5
  }
}
layer {
  name: "conv6_2_mbox_loc"
  type: "Convolution"
  bottom: "conv6_2_h"
  top: "conv6_2_mbox_loc"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 24
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv6_2_mbox_loc_perm"
  type: "Permute"
  bottom: "conv6_2_mbox_loc"
  top: "conv6_2_mbox_loc_perm"
  permute_param {
    order: 0
    order: 2
    order: 3
    order: 1
  }
}
layer {
  name: "conv6_2_mbox_loc_flat"
  type: "Reshape"
  bottom: "conv6_2_mbox_loc_perm"
  top: "conv6_2_mbox_loc_flat"
  reshape_param {
    shape {
      dim: 0
      dim: -1
      dim: 1
      dim: 1
    }
  }
}
layer {
  name: "conv6_2_mbox_conf"
  type: "Convolution"
  bottom: "conv6_2_h"
  top: "conv6_2_mbox_conf"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 12 # 126
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv6_2_mbox_conf_perm"
  type: "Permute"
  bottom: "conv6_2_mbox_conf"
  top: "conv6_2_mbox_conf_perm"
  permute_param {
    order: 0
    order: 2
    order: 3
    order: 1
  }
}
layer {
  name: "conv6_2_mbox_conf_flat"
  type: "Reshape"
  bottom: "conv6_2_mbox_conf_perm"
  top: "conv6_2_mbox_conf_flat"
  reshape_param {
    shape {
      dim: 0
      dim: -1
      dim: 1
      dim: 1
    }
  }
}
layer {
  name: "conv6_2_mbox_priorbox"
  type: "PriorBox"
  bottom: "conv6_2_h"
  bottom: "data"
  top: "conv6_2_mbox_priorbox"
  prior_box_param {
    min_size: 111.0
    max_size: 162.0
    aspect_ratio: 2
    aspect_ratio: 3
    flip: true
    clip: false
    variance: 0.1
    variance: 0.1
    variance: 0.2
    variance: 0.2
    step: 32
    offset: 0.5
  }
}
layer {
  name: "conv7_2_mbox_loc"
  type: "Convolution"
  bottom: "conv7_2_h"
  top: "conv7_2_mbox_loc"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 24
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv7_2_mbox_loc_perm"
  type: "Permute"
  bottom: "conv7_2_mbox_loc"
  top: "conv7_2_mbox_loc_perm"
  permute_param {
    order: 0
    order: 2
    order: 3
    order: 1
  }
}
layer {
  name: "conv7_2_mbox_loc_flat"
  type: "Reshape"
  bottom: "conv7_2_mbox_loc_perm"
  top: "conv7_2_mbox_loc_flat"
  reshape_param {
    shape {
      dim: 0
      dim: -1
      dim: 1
      dim: 1
    }
  }
}
layer {
  name: "conv7_2_mbox_conf"
  type: "Convolution"
  bottom: "conv7_2_h"
  top: "conv7_2_mbox_conf"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 12 # 126
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv7_2_mbox_conf_perm"
  type: "Permute"
  bottom: "conv7_2_mbox_conf"
  top: "conv7_2_mbox_conf_perm"
  permute_param {
    order: 0
    order: 2
    order: 3
    order: 1
  }
}
layer {
  name: "conv7_2_mbox_conf_flat"
  type: "Reshape"
  bottom: "conv7_2_mbox_conf_perm"
  top: "conv7_2_mbox_conf_flat"
  reshape_param {
    shape {
      dim: 0
      dim: -1
      dim: 1
      dim: 1
    }
  }
}
layer {
  name: "conv7_2_mbox_priorbox"
  type: "PriorBox"
  bottom: "conv7_2_h"
  bottom: "data"
  top: "conv7_2_mbox_priorbox"
  prior_box_param {
    min_size: 162.0
    max_size: 213.0
    aspect_ratio: 2
    aspect_ratio: 3
    flip: true
    clip: false
    variance: 0.1
    variance: 0.1
    variance: 0.2
    variance: 0.2
    step: 64
    offset: 0.5
  }
}
layer {
  name: "conv8_2_mbox_loc"
  type: "Convolution"
  bottom: "conv8_2_h"
  top: "conv8_2_mbox_loc"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 16
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv8_2_mbox_loc_perm"
  type: "Permute"
  bottom: "conv8_2_mbox_loc"
  top: "conv8_2_mbox_loc_perm"
  permute_param {
    order: 0
    order: 2
    order: 3
    order: 1
  }
}
layer {
  name: "conv8_2_mbox_loc_flat"
  type: "Reshape"
  bottom: "conv8_2_mbox_loc_perm"
  top: "conv8_2_mbox_loc_flat"
  reshape_param {
    shape {
      dim: 0
      dim: -1
      dim: 1
      dim: 1
    }
  }
}
layer {
  name: "conv8_2_mbox_conf"
  type: "Convolution"
  bottom: "conv8_2_h"
  top: "conv8_2_mbox_conf"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 8 # 84
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv8_2_mbox_conf_perm"
  type: "Permute"
  bottom: "conv8_2_mbox_conf"
  top: "conv8_2_mbox_conf_perm"
  permute_param {
    order: 0
    order: 2
    order: 3
    order: 1
  }
}
layer {
  name: "conv8_2_mbox_conf_flat"
  type: "Reshape"
  bottom: "conv8_2_mbox_conf_perm"
  top: "conv8_2_mbox_conf_flat"
  reshape_param {
    shape {
      dim: 0
      dim: -1
      dim: 1
      dim: 1
    }
  }
}
layer {
  name: "conv8_2_mbox_priorbox"
  type: "PriorBox"
  bottom: "conv8_2_h"
  bottom: "data"
  top: "conv8_2_mbox_priorbox"
  prior_box_param {
    min_size: 213.0
    max_size: 264.0
    aspect_ratio: 2
    flip: true
    clip: false
    variance: 0.1
    variance: 0.1
    variance: 0.2
    variance: 0.2
    step: 100
    offset: 0.5
  }
}
layer {
  name: "conv9_2_mbox_loc"
  type: "Convolution"
  bottom: "conv9_2_h"
  top: "conv9_2_mbox_loc"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 16
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv9_2_mbox_loc_perm"
  type: "Permute"
  bottom: "conv9_2_mbox_loc"
  top: "conv9_2_mbox_loc_perm"
  permute_param {
    order: 0
    order: 2
    order: 3
    order: 1
  }
}
layer {
  name: "conv9_2_mbox_loc_flat"
  type: "Reshape"
  bottom: "conv9_2_mbox_loc_perm"
  top: "conv9_2_mbox_loc_flat"
  reshape_param {
    shape {
      dim: 0
      dim: -1
      dim: 1
      dim: 1
    }
  }
}
layer {
  name: "conv9_2_mbox_conf"
  type: "Convolution"
  bottom: "conv9_2_h"
  top: "conv9_2_mbox_conf"
  param {
    lr_mult: 1
    decay_mult: 1
  }
  param {
    lr_mult: 2
    decay_mult: 0
  }
  convolution_param {
    num_output: 8 # 84
    pad: 1
    kernel_size: 3
    stride: 1
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
      value: 0
    }
  }
}
layer {
  name: "conv9_2_mbox_conf_perm"
  type: "Permute"
  bottom: "conv9_2_mbox_conf"
  top: "conv9_2_mbox_conf_perm"
  permute_param {
    order: 0
    order: 2
    order: 3
    order: 1
  }
}
layer {
  name: "conv9_2_mbox_conf_flat"
  type: "Reshape"
  bottom: "conv9_2_mbox_conf_perm"
  top: "conv9_2_mbox_conf_flat"
  reshape_param {
    shape {
      dim: 0
      dim: -1
      dim: 1
      dim: 1
    }
  }
}
layer {
  name: "conv9_2_mbox_priorbox"
  type: "PriorBox"
  bottom: "conv9_2_h"
  bottom: "data"
  top: "conv9_2_mbox_priorbox"
  prior_box_param {
    min_size: 264.0
    max_size: 315.0
    aspect_ratio: 2
    flip: true
    clip: false
    variance: 0.1
    variance: 0.1
    variance: 0.2
    variance: 0.2
    step: 300
    offset: 0.5
  }
}
layer {
  name: "mbox_loc"
  type: "Concat"
  bottom: "conv4_3_norm_mbox_loc_flat"
  bottom: "fc7_mbox_loc_flat"
  bottom: "conv6_2_mbox_loc_flat"
  bottom: "conv7_2_mbox_loc_flat"
  bottom: "conv8_2_mbox_loc_flat"
  bottom: "conv9_2_mbox_loc_flat"
  top: "mbox_loc"
  concat_param {
    axis: 1
  }
}
layer {
  name: "mbox_conf"
  type: "Concat"
  bottom: "conv4_3_norm_mbox_conf_flat"
  bottom: "fc7_mbox_conf_flat"
  bottom: "conv6_2_mbox_conf_flat"
  bottom: "conv7_2_mbox_conf_flat"
  bottom: "conv8_2_mbox_conf_flat"
  bottom: "conv9_2_mbox_conf_flat"
  top: "mbox_conf"
  concat_param {
    axis: 1
  }
}
layer {
  name: "mbox_priorbox"
  type: "Concat"
  bottom: "conv4_3_norm_mbox_priorbox"
  bottom: "fc7_mbox_priorbox"
  bottom: "conv6_2_mbox_priorbox"
  bottom: "conv7_2_mbox_priorbox"
  bottom: "conv8_2_mbox_priorbox"
  bottom: "conv9_2_mbox_priorbox"
  top: "mbox_priorbox"
  concat_param {
    axis: 2
  }
}

layer {
  name: "mbox_conf_reshape"
  type: "Reshape"
  bottom: "mbox_conf"
  top: "mbox_conf_reshape"
  reshape_param {
    shape {
      dim: 0
      dim: -1
      dim: 2
    }
  }
}
layer {
  name: "mbox_conf_softmax"
  type: "Softmax"
  bottom: "mbox_conf_reshape"
  top: "mbox_conf_softmax"
  softmax_param {
    axis: 2
  }
}
layer {
  name: "mbox_conf_flatten"
  type: "Reshape"
  bottom: "mbox_conf_softmax"
  top: "mbox_conf_flatten"
  reshape_param {
    shape {
      dim: 0
      dim: -1
      dim: 1
      dim: 1
    }
  }
}

layer {
  name: "detection_out"
  type: "DetectionOutput"
  bottom: "mbox_loc"
  bottom: "mbox_conf_flatten"
  bottom: "mbox_priorbox"
  top: "detection_out"
  top: "keep_count"
  include {
    phase: TEST
  }
  detection_output_param {
    num_classes: 2
    share_location: true
    background_label_id: 0
    nms_param {
      nms_threshold: 0.3
      top_k: 400
    }
    code_type: CENTER_SIZE
    keep_top_k: 200
    confidence_threshold: 0.01
  }
}

Thanks.

Thanks for the reply,

I used your deploy.prototxt and I correctly converted in TensorRT, but unfortunately with a simple detector program like your “DeepStream 4.2.1 Exercise” the output:

0:05:22.716828271 25799 0x55895d0000 ERROR nvinfer gstnvinfer.cpp:511:gst_nvinfer_logger: NvDsInferContext[UID 1]:parseBoundingBox(): Could not find output coverage layer for parsing objects
0:05:22.716895668 25799 0x55895d0000 ERROR nvinfer gstnvinfer.cpp:511:gst_nvinfer_logger: NvDsInferContext[UID 1]:fillDetectionOutput(): Failed to parse bboxes
Segmentation fault (core dumped)

how can I solve this?

Best Regards,
Andrea

Hi,

The output layer name change in the this model.
Please update it correctly.

Thanks.

Hi,

Did you train your model using resnet in caffe?? or VGG 16 in caffe??

I wanted to train my model in caffe to run inference. But I want to know which repo is best for TRAINING OWN DATASET.