Amazon Fulfillment Bin Imagery Dataset

Images and associated metadata are derived from the operational environment of an Amazon Fulfillment Centre. The data captures the contents of bins within a pod structure, specifically recorded as robotic units execute their normal transportation duties. This collection provides 536,434 images along with accompanying tabular metadata, designed primarily to offer file locations (S3 URIs) and precise counts of items observed within each bin image. This resource is valuable for analysis and enabling distributed learning applications in logistics technology.

The dataset features a high usability rating of 10.00 and is provided as a static resource, with an expected update frequency of 'Never'. The main file list is provided as a 21.94 MB CSV file, titled quantity.csv, detailing the image metadata. The overall collection contains 536,434 records. The images themselves follow three naming conventions: files named with 1 to 4 digits (1,200 files), 5 digits (99,999 files), or 6 digits (435,235 files), ranging up to 535234.jpg.

This data product is suited for applications involving computer vision, classification, and developing machine learning models related to object detection and counting. Specific use cases include training algorithms for inventory validation, assessing product density in fulfillment bins, and automating visual inspection processes within warehouse settings. The file list is structured to support distributed learning systems effectively.

The data reflects bin imagery captured during routine operations within a specific Amazon Fulfillment Centre. The data scope is focused on visual imagery of bin contents and associated item counts during robotic transport of pods. The dataset is provided in a final state and is not expected to receive further updates.

CC BY-NC-SA 4.0

Machine learning specialists and data scientists who require large, labelled image datasets for training computer vision models. Researchers focusing on automation, robotics, and logistical efficiency within modern warehousing environments. Academics exploring complex classification tasks or item density analysis.

Original Data Source: Amazon Fulfillment Bin Imagery Dataset