Semantic segmentation pretrained caffe model

I am trying to perform semantic segmentation inference on Jetson nano using a pretrained caffe model. When using the segnet-console example to convert the model into a GIE representation I get error for using the below network layers:

layer { bottom: ‘conv1_1’ top: ‘conv1_1’ name: ‘bn1_1’ type: “BN”
bn_param { bn_mode: INFERENCE scale_filler { type: ‘constant’ value: 1 }
shift_filler { type: ‘constant’ value: 0.001 } } }

layer { type: “UnPooling” bottom: “fc6-deconv” bottom: “pool5_mask” top: “unpool5” name: “unpool5”
pooling_param { pool: MAX kernel_size: 2 stride: 2 }

Is there any way of converting them, or possibly bypassing them. I think there is no BN implementation in tensorrt, and when using BatchNorm i get the following error(As it is a different type of layer and the weight do not match any more):

CaffeParser: ERROR: Attempting to access NULL weights
segnet-console: caffe/caffeParser.cpp:1000: nvinfer1::ILayer* parseBatchNormalization(nvinfer1::INetworkDefinition&, const ditcaffe::LayerParameter&, CaffeWeightFactory&, BlobNameToTensor&): Assertion `mean.count == variance.count && movingAverage.count == 1’ failed.

Any help would be highly appreciated. Thank you!

I should have mentioned that I would like to keep and use the pretraiend weights as I do not have access to the training dataset.