VIC Configuration failed image scale factor exceeds 16

[a9010028]

Hi Sir: I find an issue is about Jetson platform problem.

==================================================================
/dvs/git/dirty/git-master_linux/nvutils/nvbufsurftransform/nvbufsurftransform.cpp:4726: => VIC Configuration failed image scale factor exceeds 16, use GPU for Transformation
1:11:01.608351383 2018863 0xaaaafe3e21e0 WARN nvinfer gstnvinfer.cpp:1477:convert_batch_and_push_to_input_thread: error: NvBufSurfTransform failed with error -3 while converting buffer
1:11:01.608493595 2018863 0xaaaafe3e21e0 WARN nvinfer gstnvinfer.cpp:2420:gst_nvinfer_output_loop: error: Internal data stream error.
1:11:01.608521243 2018863 0xaaaafe3e21e0 WARN nvinfer gstnvinfer.cpp:2420:gst_nvinfer_output_loop: error: streaming stopped, reason error (-5)
ERROR from element secondary-infer-engine1: NvBufSurfTransform failed with error -3 while converting buffer
Error details: /dvs/git/dirty/git-master_linux/deepstream/sdk/src/gst-plugins/gst-nvinfer/gstnvinfer.cpp(1477): convert_batch_and_push_to_input_thread (): /GstPipeline:ds-lpr-pipeline/GstNvInfer:secondary-infer-engine1

When i using the 416*416 , maybe make the output can't be alignment 16.

==================================================================
0:00:10.583964075 2018863 0xaaaafe9d78a0 INFO nvinfer gstnvinfer.cpp:682:gst_nvinfer_logger: NvDsInferContext[UID 2]: Info from NvDsInferContextImpl::deserializeEngineAndBackend() <nvdsinfer_context_impl.cpp:2095> [UID = 2]: deserialized trt engine from :/home/aaeon/ds/weights/tlpr.onnx_b1_gpu0_fp16.engine
WARNING: [TRT]: The getMaxBatchSize() function should not be used with an engine built from a network created with NetworkDefinitionCreationFlag::kEXPLICIT_BATCH flag. This function will always return 1.
INFO: [Implicit Engine Info]: layers num: 4
0 INPUT kFLOAT input 3x416x416
1 OUTPUT kFLOAT boxes 3549x4
2 OUTPUT kFLOAT scores 3549x1
3 OUTPUT kFLOAT classes 3549x1

How can i fix this problem.
thank you

[nkinnaird]

The Jetson can't scale by a factor more than 16 when resizing frames for input into model. Either frames upstream need to be resized or model input size needs to be larger. Former can be done with some kind of capsfilter in a Gstreamer pipeline, like :

! video/x-raw(memory:NVMM),width=1280,height=720 !

[marcoslucianops]

You can set scaling-compute-hw=1 in the [property] key on the config_infer file.