Multi-Processing and Shared caching based Quality Assurance system for supply chain of fruit delivery and import/export.
During cleaning and preprocessing, we removed the images with labels 'image_quality' and 'condition' as we only wanted to train on high quality data. Also, we would not be using 'artifact' label in training but we will keep the images, and 'pedicel' label would go into segmentation model but not in post processing.
Processed images and masks are stored as numpy arrays in npy format. Processed masks have 6 channels (equal to number of classes defined in labels.json) and ordered in ascending order of values (category_ids)
Abhimanyu Banerjee CSE, Bennett University Gurgaon, India Ab8963@benett.edu.in
Anudit Nagar CSE, Bennett University Noida, India An9316@bennett.edu.in
Himanshu Mittal CSE, Bennett University Bahadurgarh, India Hm6729@bennett.edu.in