How to run tlt-converter

Accelerated Computing Intelligent Video Analytics TAO Toolkit
1

$ wget https://developer.nvidia.com/tlt-converter-trt71

Note, please download tlt-converter with dla enabled. wget https://developer.nvidia.com/assets/TLT/Secure/tlt-converter-7.1-dla.zip

$ unzip tlt-converter-trt71

$ chmod +x tlt-converter

For FaceDetectIR (https://ngc.nvidia.com/catalog/models/nvidia:tlt_facedetectir)

$ wget https://api.ngc.nvidia.com/v2/models/nvidia/tlt_facedetectir/versions/pruned_v1.0/files/resnet18_facedetectir_pruned.etlt

$ wget https://api.ngc.nvidia.com/v2/models/nvidia/tlt_facedetectir/versions/pruned_v1.0/files/facedetectir_int8.txt

$ ./tlt-converter resnet18_facedetectir_pruned.etlt -k tlt_encode -c facedetectir_int8.txt -o output_cov/Sigmoid,output_bbox/BiasAdd -d 3,240,384 -i nchw -e facedetectir_int8.engine -m $MAX_BATCH_SIZE -t $INFERENCE_PRECISION -b $BATCH_SIZE

For example,
$ ./tlt-converter resnet18_facedetectir_pruned.etlt -k tlt_encode -c facedetectir_int8.txt -o output_cov/Sigmoid,output_bbox/BiasAdd -d 3,240,384 -i nchw -m 64 -t int8 -e facedetectir_int8.engine -b 64

For TrafficCamNet (https://ngc.nvidia.com/catalog/models/nvidia:tlt_trafficcamnet)

$ wget https://api.ngc.nvidia.com/v2/models/nvidia/tlt_trafficcamnet/versions/pruned_v1.0/files/resnet18_trafficcamnet_pruned.etlt

$ wget https://api.ngc.nvidia.com/v2/models/nvidia/tlt_trafficcamnet/versions/pruned_v1.0/files/trafficnet_int8.txt

$ ./tlt-converter resnet18_trafficcamnet_pruned.etlt -k tlt_encode -c trafficnet_int8.txt -o output_cov/Sigmoid,output_bbox/BiasAdd -d 3,544,960 -i nchw -e trafficnet_int8.engine -m $MAX_BATCH_SIZE -t $INFERENCE_PRECISION -b $BATCH_SIZE

For example,

$ ./tlt-converter resnet18_trafficcamnet_pruned.etlt -k tlt_encode -c trafficnet_int8.txt -o output_cov/Sigmoid,output_bbox/BiasAdd -d 3,544,960 -i nchw -e trafficnet_int8.engine -m 64 -t int8 -b 64

For PeopleNet (https://ngc.nvidia.com/catalog/models/nvidia:tlt_peoplenet)

$ wget https://api.ngc.nvidia.com/v2/models/nvidia/tlt_peoplenet/versions/pruned_v2.0/files/resnet18_peoplenet_pruned.etlt

$ wget https://api.ngc.nvidia.com/v2/models/nvidia/tlt_peoplenet/versions/pruned_v2.0/files/resnet18_peoplenet_int8.txt

$ ./tlt-converter resnet18_peoplenet_pruned.etlt -k tlt_encode -c resnet18_peoplenet_int8.txt -o output_cov/Sigmoid,output_bbox/BiasAdd -d 3,544,960 -i nchw -e peoplenet_int8.engine -m $MAX_BATCH_SIZE -t $INFERENCE_PRECISION -b $BATCH_SIZE

For example,

$ ./tlt-converter resnet18_peoplenet_pruned.etlt -k tlt_encode -c resnet18_peoplenet_int8.txt -o output_cov/Sigmoid,output_bbox/BiasAdd -d 3,544,960 -i nchw -e peoplenet_int8.engine -m 64 -t int8 -b 64

For DashCamNet (https://ngc.nvidia.com/catalog/models/nvidia:tlt_dashcamnet)

$ wget https://api.ngc.nvidia.com/v2/models/nvidia/tlt_dashcamnet/versions/pruned_v1.0/files/resnet18_dashcamnet_pruned.etlt

$ wget https://api.ngc.nvidia.com/v2/models/nvidia/tlt_dashcamnet/versions/pruned_v1.0/files/dashcamnet_int8.txt

$ ./tlt-converter resnet18_dashcamnet_pruned.etlt -k tlt_encode -c dashcamnet_int8.txt -o output_cov/Sigmoid,output_bbox/BiasAdd -d 3,544,960 -i nchw -e dashcam_int8.engine -m $MAX_BATCH_SIZE -t $INFERENCE_PRECISION -b $BATCH_SIZE

For example,

$ ./tlt-converter resnet18_dashcamnet_pruned.etlt -k tlt_encode -c dashcamnet_int8.txt -o output_cov/Sigmoid,output_bbox/BiasAdd -d 3,544,960 -i nchw -e dashcam_int8.engine -m 64 -t int8 -b 64

For VehicleMakeNet (https://ngc.nvidia.com/catalog/models/nvidia:tlt_vehiclemakenet)

$ wget https://api.ngc.nvidia.com/v2/models/nvidia/tlt_vehiclemakenet/versions/pruned_v1.0/files/resnet18_vehiclemakenet_pruned.etlt

$ wget https://api.ngc.nvidia.com/v2/models/nvidia/tlt_vehiclemakenet/versions/pruned_v1.0/files/vehiclemakenet_int8.txt

$ ./tlt-converter resnet18_vehiclemakenet_pruned.etlt -k tlt_encode -c vehiclemakenet_int8.txt -o predictions/Softmax -d 3,224,224 -i nchw -e vehiclemakenet_int8.engine -m $MAX_BATCH_SIZE -t $INFERENCE_PRECISION -b $BATCH_SIZE

For example,

$ ./tlt-converter resnet18_vehiclemakenet_pruned.etlt -k tlt_encode -c vehiclemakenet_int8.txt -o predictions/Softmax -d 3,224,224 -i nchw -e vehiclemakenet_int8.engine -m 64 -t int8 -b 64

For VehicleTypeNet (https://ngc.nvidia.com/catalog/models/nvidia:tlt_vehicletypenet)

$ wget https://api.ngc.nvidia.com/v2/models/nvidia/tlt_vehicletypenet/versions/pruned_v1.0/files/resnet18_vehicletypenet_pruned.etlt

$ wget https://api.ngc.nvidia.com/v2/models/nvidia/tlt_vehicletypenet/versions/pruned_v1.0/files/vehicletypenet_int8.txt

$ ./tlt-converter resnet18_vehicletypenet_pruned.etlt -k tlt_encode -c vehicletypenet_int8.txt -o predictions/Softmax -d 3,224,224 -i nchw -e vehicletypenet_int8.engine -m $MAX_BATCH_SIZE -t $INFERENCE_PRECISION -b $BATCH_SIZE

For example,

$ ./tlt-converter resnet18_vehicletypenet_pruned.etlt -k tlt_encode -c vehicletypenet_int8.txt -o predictions/Softmax -d 3,224,224 -i nchw -e vehicletypenet_int8.engine -m 64 -t int8 -b 64

2 Likes