Trying to install nvOCDR

I tried installing Nvidia Optical Character Detection and Recognition but when I tried to generate the TensorRT engine it has an error.

/usr/src/tensorrt/bin/trtexec --onnx=./ocdnet.onnx --minShapes=input:1x3x736x1280 --optShapes=input:1x3x736x1280 --maxShapes=input:4x3x736x1280 --fp16 --saveEngine=./ocdnet.fp16.engine

This is the error when running the command

[04/02/2024-21:36:50] [I] === Model Options ===
[04/02/2024-21:36:50] [I] Format: ONNX
[04/02/2024-21:36:50] [I] Model: ./ocdnet.onnx
[04/02/2024-21:36:50] [I] Output:
[04/02/2024-21:36:50] [I] === Build Options ===
[04/02/2024-21:36:50] [I] Max batch: explicit batch
[04/02/2024-21:36:50] [I] Memory Pools: workspace: default, dlaSRAM: default, dlaLocalDRAM: default, dlaGlobalDRAM: default
[04/02/2024-21:36:50] [I] minTiming: 1
[04/02/2024-21:36:50] [I] avgTiming: 8
[04/02/2024-21:36:50] [I] Precision: FP32+FP16
[04/02/2024-21:36:50] [I] LayerPrecisions:
[04/02/2024-21:36:50] [I] Calibration:
[04/02/2024-21:36:50] [I] Refit: Disabled
[04/02/2024-21:36:50] [I] Sparsity: Disabled
[04/02/2024-21:36:50] [I] Safe mode: Disabled
[04/02/2024-21:36:50] [I] DirectIO mode: Disabled
[04/02/2024-21:36:50] [I] Restricted mode: Disabled
[04/02/2024-21:36:50] [I] Build only: Disabled
[04/02/2024-21:36:50] [I] Save engine: ./ocdnet.fp16.engine
[04/02/2024-21:36:50] [I] Load engine:
[04/02/2024-21:36:50] [I] Profiling verbosity: 0
[04/02/2024-21:36:50] [I] Tactic sources: Using default tactic sources
[04/02/2024-21:36:50] [I] timingCacheMode: local
[04/02/2024-21:36:50] [I] timingCacheFile:
[04/02/2024-21:36:50] [I] Heuristic: Disabled
[04/02/2024-21:36:50] [I] Preview Features: Use default preview flags.
[04/02/2024-21:36:50] [I] Input(s)s format: fp32:CHW
[04/02/2024-21:36:50] [I] Output(s)s format: fp32:CHW
[04/02/2024-21:36:50] [I] Input build shape: input=1x3x736x1280+1x3x736x1280+4x3x736x1280
[04/02/2024-21:36:50] [I] Input calibration shapes: model
[04/02/2024-21:36:50] [I] === System Options ===
[04/02/2024-21:36:50] [I] Device: 0
[04/02/2024-21:36:50] [I] DLACore:
[04/02/2024-21:36:50] [I] Plugins:
[04/02/2024-21:36:50] [I] === Inference Options ===
[04/02/2024-21:36:50] [I] Batch: Explicit
[04/02/2024-21:36:50] [I] Input inference shape: input=1x3x736x1280
[04/02/2024-21:36:50] [I] Iterations: 10
[04/02/2024-21:36:50] [I] Duration: 3s (+ 200ms warm up)
[04/02/2024-21:36:50] [I] Sleep time: 0ms
[04/02/2024-21:36:50] [I] Idle time: 0ms
[04/02/2024-21:36:50] [I] Streams: 1
[04/02/2024-21:36:50] [I] ExposeDMA: Disabled
[04/02/2024-21:36:50] [I] Data transfers: Enabled
[04/02/2024-21:36:50] [I] Spin-wait: Disabled
[04/02/2024-21:36:50] [I] Multithreading: Disabled
[04/02/2024-21:36:50] [I] CUDA Graph: Disabled
[04/02/2024-21:36:50] [I] Separate profiling: Disabled
[04/02/2024-21:36:50] [I] Time Deserialize: Disabled
[04/02/2024-21:36:50] [I] Time Refit: Disabled
[04/02/2024-21:36:50] [I] NVTX verbosity: 0
[04/02/2024-21:36:50] [I] Persistent Cache Ratio: 0
[04/02/2024-21:36:50] [I] Inputs:
[04/02/2024-21:36:50] [I] === Reporting Options ===
[04/02/2024-21:36:50] [I] Verbose: Disabled
[04/02/2024-21:36:50] [I] Averages: 10 inferences
[04/02/2024-21:36:50] [I] Percentiles: 90,95,99
[04/02/2024-21:36:50] [I] Dump refittable layers:Disabled
[04/02/2024-21:36:50] [I] Dump output: Disabled
[04/02/2024-21:36:50] [I] Profile: Disabled
[04/02/2024-21:36:50] [I] Export timing to JSON file:
[04/02/2024-21:36:50] [I] Export output to JSON file:
[04/02/2024-21:36:50] [I] Export profile to JSON file:
[04/02/2024-21:36:50] [I]
Cuda failure: initialization error
Aborted (core dumped)

Thank you in advance for your assistance.

Hi,

Cuda failure: initialization error

The error is environment setting related.
Could you try the below command to verify TensorRT functionality first:

$ /usr/src/tensorrt/bin/trtexec --onnx=/usr/src/tensorrt/data/resnet50/ResNet50.onnx

Thanks.