Invalid argument: warmup setting expects n bytes

Description

Hi, I am trying to set up some warm up input files for my TensorFlow model, but I am running into the following error:

E0715 02:08:56.141822 1 model_repository_manager.cc:970] failed to load 'efficientdet_graphdef' version 3: Invalid argument: warmup setting expects 6220800 bytes, but the data provided from img.pngonly has 4080549 bytes

Is it necessary for the file size of the image to be the same as the product of the dims specified in the model config?

This is the image I am trying to send for warm up: https://user-images.githubusercontent.com/11736571/77320690-099af300-6d37-11ea-9d86-24f14dc2d540.png

Triton Information

I am using Triton v2.0.0. I built the Triton container locally (only change was setting the Triton minimum compute capability to 3.7).

Model Config

name: "efficientdet_graphdef"
platform: "tensorflow_graphdef"
max_batch_size: 1
input [
    {
        name: "image_arrays"
        data_type: TYPE_UINT8
        format: FORMAT_NHWC
        dims: [ -1, -1, 3 ]
    }
]
output [
    {
        name: "detections"
        data_type: TYPE_FP32
        dims: [ -1, 7 ]
    }
]
instance_group [
    {
        kind: KIND_GPU
        count: 1
        gpus: [ 0 ]
    }
]
model_warmup [
    {
        name: "warmup_file"
        batch_size: 1
        inputs: {
            key: "input"
            value: {
                data_type: TYPE_UINT8
                dims: [ 1080, 1920, 3 ]
                input_data_file: "img.png"
            }
        }
    }
]

[GuanLuo]

@masguit42 What is the error message that you received? And do you mind to create a separate issue for your questions?

[GuanLuo]

As stated here, the file must contain raw data for the inputs. Base on the config you provided, it is data for UINT8 matrix with shape [1080, 1920, 3], but the file you provided is a png file, you should preprocess your image to the desired form