Mismatched Number of Channels for YOLOv3 model

[Sokna_Ly]

I am currently working on a deep learning project using the arcgis.learn module in the ArcGIS API for Python. The goal of the project is to utilize the AutoDL class for automated model selection and training for object detection tasks. The training dataset comprises TIFF format images with PASCAL VOC annotation, and the images are of size 448x448 pixels.

While using AutoDL to train object detection models, I encountered a runtime error specifically related to the YOLOv3 model:

RuntimeError: Given groups=1, weight of size [32, 4, 3, 3], expected input[2, 3, 224, 224] to have 4 channels, but got 3 channels instead

I have 2 questions regarding this error.

1. How should i modify my input training data to match the number of channels required by this model?
2. Are there ways I can exclude YOLOv3 (or any models) from AutoDL? 

I appreciate any insights, suggestions, or alternative approaches from the community to address these challenges. Thank you for your valuable input!

[PavanYadav]

@Sokna_Ly 

try the advance option and select desired model types 

@PriyankaTuteja - can you respond to #1. 

Cheers!

Pava

[PriyankaTuteja]

The arcgis.learn api has autodl class that provides a parameter named: network, which allows you to enter a a list of networks you'd like to evaluate.  

@Sokna_Ly you could pass in networks, and omit yolov3 from this.

Additionally, in case you want us to investigate the issue with yolov3, please reach out to me @ptuteja@esri.com