BDryFish: A smartphone image dataset of dried fish (Shutki) trade categories from Bangladeshi markets
Data in Brief · 2026-08 · 69 · 113359
DOI: 10.1016/j.dib.2026.113359 ↗Kurz gefasst
Fachartikel in der Zeitschrift Data in Brief, 2026-08. Verfasst von Tanvir Anjum Apurbo und 4 weiteren Autor:innen. Beteiligte Einrichtung: Department of Computer Science & Engineering, Southeast University, 252, Tejgaon Industrial Area, Dhaka, 1208, Bangladesh. Die Arbeit nennt 12 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Dried fish, known in Bangladesh as shutki, is traded under local names that do not correspond neatly to biological species. BDryFish contains 5778 photographs of 22 dried fish trade categories, taken with three smartphones, with and without flash, on 14 and 21 May 2025 at two wholesale markets in Dhaka, Karwan Bazar and Rayerbazar Shutki Arot. Every category is shown as isolated specimens (Single, 2981 images) and as market heaps (Bulk, 2797 images). The labels are the Bangla trade names given by sellers, so the label space combines species with processing state and cut. English and scientific names were matched from online sources and are not independent taxonomic identifications. Two pairs of classes share a species (whole and sliced hilsa, sun dried and fermented swamp barb), two hairtail classes are size grades, Chouka has no established scientific identity, and one class is a crustacean. Images are released at full resolution and at 640 pixels on the long side, without padding. A metadata table records capture time, device, lens, market, flash state and capture session identifiers at four time thresholds. Because frames were taken in bursts, a random split would put near identical frames in training and test sets, so the recommended split keeps whole capture sessions together. The dataset supports image classification benchmarks under market conditions, transfer between isolated and heaped presentations, shifts across devices and capture days, and recognition of processing states for food authentication. It is an image classification resource; object detection would require additional annotation.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Computer Science & Engineering, Southeast University, 252, Tejgaon Industrial Area, Dhaka, 1208, Bangladesh.
Zitieren
APA
Apurbo, T. A., Ahammed, K. F., Abdullah, M., Islam, M., & Belali, M. H. (2026). BDryFish: A smartphone image dataset of dried fish (Shutki) trade categories from Bangladeshi markets. Data in Brief, 69, 113359. https://doi.org/10.1016/j.dib.2026.113359
Vancouver
Apurbo TA, Ahammed KF, Abdullah M, Islam M, Belali MH. BDryFish: A smartphone image dataset of dried fish (Shutki) trade categories from Bangladeshi markets. Data in Brief. 2026;69:113359. doi:10.1016/j.dib.2026.113359
BibTeX
@article{Apurbo2026,
author = {Apurbo, Tanvir Anjum and Ahammed, Kh Faysal and Abdullah, Md and Islam, Mohosinin and Belali, Md Habibullah},
title = {BDryFish: A smartphone image dataset of dried fish (Shutki) trade categories from Bangladeshi markets},
journal = {Data in Brief},
year = {2026},
volume = {69},
pages = {113359},
doi = {10.1016/j.dib.2026.113359}
}
EUVIMED