@siddgood @aws-dash @pooyavahidi pooyavahidi I used the scripts for data preprocessing and evaluation. I just changed the S3 bucket name to binary image data. It was updated in the code. The files generated on the S3 bucket are empty. I followed the steps mentioned in the readme and created the custom folder, uploading the pizza/non-pizza images separately in the folder.

preprocessing.py

 DATASET_NAME="binaryimagedata" BASE_DIR="/opt/ml/processing/input/data" IM2REC_PATH="/opt/ml/processing/input/code/im2rec.py" 

evalution.py

 ###### 2. Loading and preparing test .rec file ###### logger.debug("Reading test data.") DATASET_NAME="binaryimagedata" test_path = f"/opt/ml/processing/test/{DATASET_NAME}_test.rec" 

ClientError: ClientError: Unable to start training. Empty .rec file in train channel., exit code: 2