sosiklab/NES-plankton-classifier-2022-dataset

NES Plankton Classifier 2022 Training Data

Image classification dataset of plankton species imaged by Imaging FlowCytobot (IFCB) on the Northeast US Shelf. Used to train the NES Plankton Classifier 2022.

Dataset Summary

• 155 classes of plankton and non-plankton ROIs (including detritus, bubbles, fibers, etc.)
• 97,026 labeled images across train and validation splits
• Images are grayscale PNGs extracted from IFCB sample files
• Annotations are human-verified using the IFCB Annotate tool

Years

ROIs span 2006–2021, unevenly distributed across years. Click to expand

Year	ROIs
2006	5,418
2007	7,731
2008	8,426
2009	10,262
2010	6,945
2011	7,302
2012	8,209
2013	7,486
2014	5,888
2015	6,197
2016	6,244
2017	7,146
2018	2,292
2019	7,403
2020	29
2021	47

ifcb_roi_pid appears in two formats, which are instrument-dependent rather than strictly year-dependent:

• Legacy (IFCB<n>_YYYY_DDD_HHMMSS_NNNNN) — appears in 2006–2017 data
• Modern (DYYYYMMDDThhmmss_IFCB<nnn>_NNNNN) — appears in 2012–2021 data

Both formats coexist in 2012–2017, since different instruments switched over at different times.

Usage

Here's the easiest way to access parquet data.

from datasets import load_dataset

ds = load_dataset("sosiklab/NES-plankton-classifier-2022-dataset")

example = ds["train"][0]
example["image"]        # PIL.Image.Image
example["label"]        # int class index
example["classname"]    # human-readable class name
example["ifcb_roi_pid"] # IFCB ROI persistent identifier

# Map int label -> name via ClassLabel metadata
ds["train"].features["label"].int2str(example["label"])

Exporting Parquet to Filesystem Images and Tabular CSV

Sometimes it's easier to work directly in the filesystem and simple dataformats. Use the following code to convert this repo's dataset into png images and CSV tabular data.

from pathlib import Path
import csv
from datasets import load_dataset


# Export Functions #

def save_images(ds, output_dir, split_subfolders=False, classname_subfolders=False, clobber=False):
    paths = {}
    for split_name, dset in ds.items():
        for ex in dset:

            path = Path(output_dir)
            if split_subfolders:
                path = path / split_name
            if classname_subfolders:
                path = path / ex["classname"]
            path.mkdir(parents=True, exist_ok=True)

            path = path / f'{ex["ifcb_roi_pid"]}.png'
            if clobber or not path.exists():
                ex["image"].save(path)

            paths[ex["ifcb_roi_pid"]] = str(path)

    return paths


def save_table_csv(ds, csv_path):
    csv_path = Path(csv_path)
    csv_path.parent.mkdir(parents=True, exist_ok=True)

    with csv_path.open("w", newline="") as f:
        writer = csv.writer(f)
        writer.writerow(["ifcb_roi_pid", "dataset", "class_label", "class_name"])

        for split, dset in ds.items():
            for ex in dset:
                writer.writerow([
                    ex["ifcb_roi_pid"],
                    split,
                    ex["label"],
                    ex["classname"],
                ])

    return str(csv_path)


# Run Dataset png and csv Export functions #

ds = load_dataset("sosiklab/NES-plankton-classifier-2022-dataset")

image_paths = save_images(
    ds,
    "image_data/NES-plankton-classifier-2022-dataset",
    split_subfolders=True,
    classname_subfolders=True,
)

csv_path = save_table_csv(ds, "NES-plankton-classifier-2022-dataset.csv")

Columns

Column	Type	Description
image	Image	Grayscale ROI image from IFCB
label	ClassLabel	Integer class index
classname	string	Human-readable class name (same as label decoded)
ifcb_roi_pid	string	IFCB ROI persistent identifier

Splits

Split	Examples
train	77,617
validation	19,409

The 80/20 split was made per class (stratified), not per IFCB bin or fully random across all ROIs.

Per-class train/validation counts and ratio. Click to expand

Class	Total	Train	Validation	Ratio (train:val)
Acantharia	53	42	11	79:21
Acanthoica_quattrospina	854	683	171	80:20
Akashiwo	31	25	6	81:19
Alexandrium_catenella	319	255	64	80:20
Amoeba	1008	806	202	80:20
Amphidinium	1045	836	209	80:20
Amylax	437	350	87	80:20
Apedinella	344	275	69	80:20
Asterionellopsis_glacialis	1125	900	225	80:20
Bacillaria	40	32	8	80:20
Bacillariophyceae	212	170	42	80:20
Bacillariophyceae_morphotype1	746	597	149	80:20
Bacteriastrum	34	27	7	79:21
Balanion	1106	885	221	80:20
Biddulphia	46	37	9	80:20
Calciopappus	1286	1029	257	80:20
Calciosolenia_brasiliensis	28	22	6	79:21
Cerataulina_pelagica	949	759	190	80:20
Chaetoceros	936	749	187	80:20
Chaetoceros_danicus	677	542	135	80:20
Chaetoceros_didymus	637	510	127	80:20
Chaetoceros_didymus_TAG_external_flagellate	1168	934	234	80:20
Chaetoceros_peruvianis	30	24	6	80:20
Chaetoceros_similis	554	443	111	80:20
Chaetoceros_socialis	215	172	43	80:20
Chaetoceros_subtilis	199	159	40	80:20
Chaetoceros_tenuissimus	242	194	48	80:20
Chaetoceros_throndsenii	54	43	11	80:20
Chrysochromulina	1134	907	227	80:20
Chrysochromulina_lanceolata	602	482	120	80:20
Ciliophora	1282	1026	256	80:20
Copepod_nauplii	155	124	31	80:20
Corethron_hystrix	1170	936	234	80:20
Corymbellus	314	251	63	80:20
Coscinodiscus	754	603	151	80:20
Cryptophyta	1111	889	222	80:20
Cylindrotheca	509	407	102	80:20
Cylindrotheca_morphotype1	485	388	97	80:20
Dactyliosolen_blavyanus	1158	926	232	80:20
Dactyliosolen_fragilissimus	875	700	175	80:20
Delphineis	528	422	106	80:20
Dictyocha	1194	955	239	80:20
Dictyocysta	31	25	6	81:19
Didinium	57	46	11	81:19
Dinobryon	1203	962	241	80:20
Dinophyceae	1089	871	218	80:20
Dinophysis_acuminata	181	145	36	80:20
Dinophysis_norvegica	482	386	96	80:20
Dinophysis_tripos	28	22	6	79:21
Ditylum_brightwellii	1181	945	236	80:20
Emiliania_huxleyi	186	149	37	80:20
Ephemera	715	572	143	80:20
Eucampia	1586	1269	317	80:20
Eucampia_morphytype1	413	330	83	80:20
Euglena	930	744	186	80:20
Euplotes	54	43	11	80:20
Euplotes_morphotype1	66	53	13	80:20
Eutintinnus	98	78	20	80:20
Favella	208	166	42	80:20
Gonyaulax	892	714	178	80:20
Guinardia_delicatula	2000	1600	400	80:20
Guinardia_delicatula_TAG_internal_parasite	1143	914	229	80:20
Guinardia_flaccida	1079	863	216	80:20
Guinardia_striata	1029	823	206	80:20
Gyrodinium	1130	904	226	80:20
Hemiaulus	1146	917	229	80:20
Hemiaulus_membranaceus	40	32	8	80:20
Heterocapsa_rotundata	1050	840	210	80:20
Karenia	664	531	133	80:20
Katodinium_or_Torodinium	1134	907	227	80:20
Kryptoperidinium_triquetrum	1866	1493	373	80:20
Laboea_strobila	1131	905	226	80:20
Lauderia_annulata	245	196	49	80:20
Leegaardiella_ovalis	474	379	95	80:20
Leptocylindrus	1140	912	228	80:20
Leptocylindrus_mediterraneus	476	381	95	80:20
Licmophora	379	303	76	80:20
Margalefidinium	1760	1408	352	80:20
Mesodinium	1152	922	230	80:20
Nanoneis	135	108	27	80:20
Odontella	93	74	19	80:20
Ophiaster	76	61	15	80:20
Oxytoxum	960	768	192	80:20
Paralia_sulcata	730	584	146	80:20
Paratontonia_gracillima	1177	942	235	80:20
Parvicorbicula_socialis	160	128	32	80:20
Pelagostrobilidium	504	403	101	80:20
Phaeocystis	1165	932	233	80:20
Phaeocystis_debris	242	194	48	80:20
Pleuronema	221	177	44	80:20
Pleurosigma	1181	945	236	80:20
Polykrikos	129	103	26	80:20
Prorocentrum	795	636	159	80:20
Prorocentrum_dentatum	683	546	137	80:20
Prorocentrum_micans	883	706	177	80:20
Prorocentrum_triestinum	1188	950	238	80:20
Proterythropsis	767	614	153	80:20
Protoperidinium	173	138	35	80:20
Pseudo-nitzschia	1153	922	231	80:20
Pseudochattonella_farcimen	1588	1270	318	80:20
Pyramimonas	1113	890	223	80:20
Pyramimonas_longicauda	1186	949	237	80:20
Pyramimonas_morphotype1	93	74	19	80:20
Rhabdolithes	48	38	10	79:21
Rhizosolenia	1137	910	227	80:20
Scrippsiella	116	93	23	80:20
Skeletonema	1130	904	226	80:20
Stenosemella_morphotype1	103	82	21	80:20
Stenosemella_pacifica	1150	920	230	80:20
Stephanopyxis	86	69	17	80:20
Strombidium_capitatum	79	63	16	80:20
Strombidium_conicum	143	114	29	80:20
Strombidium_inclinatum	321	257	64	80:20
Strombidium_morphotype1	1239	991	248	80:20
Strombidium_morphotype2	487	390	97	80:20
Strombidium_tintinnodes	366	293	73	80:20
Strombidium_wulffi	235	188	47	80:20
Syracosphaera_pulchra	463	370	93	80:20
Thalassionema	1163	930	233	80:20
Thalassiosira	1309	1047	262	80:20
Thalassiosira_TAG_external_detritus	1089	871	218	80:20
Thalassiosira_sp_aff_mala	604	483	121	80:20
Tiarina_fusus	57	46	11	81:19
Tintinnidium_mucicola	29	23	6	79:21
Tintinnina	245	196	49	80:20
Tintinnopsis	1109	887	222	80:20
Tontonia_appendiculariformis	180	144	36	80:20
Trichodesmium	51	41	10	80:20
Tripos	149	119	30	80:20
Tripos_furca	342	274	68	80:20
Tripos_fusus	63	50	13	79:21
Tripos_lineatus	945	756	189	80:20
Vicicitus_globosus	445	356	89	80:20
Warnowia	73	58	15	79:21
bead	1122	898	224	80:20
bubble	34	27	7	79:21
camera_spot	494	395	99	80:20
coccolithophorid	302	242	60	80:20
detritus	1017	814	203	80:20
detritus_transparent	761	609	152	80:20
fecal_pellet	219	175	44	80:20
fiber	1145	916	229	80:20
fiber_TAG_external_detritus	219	175	44	80:20
flagellate	962	770	192	80:20
flagellate_morphotype1	1170	936	234	80:20
flagellate_morphotype3	1124	899	225	80:20
nanoplankton_mix	1397	1118	279	80:20
pennate	905	724	181	80:20
pennate_Pseudo-nitzschia	194	155	39	80:20
pennate_Thalassionema	45	36	9	80:20
pennate_morphotype1	287	230	57	80:20
pollen	41	33	8	80:20
shellfish_larvae	31	25	6	81:19
unknown2	58	46	12	79:21
zooplankton	64	51	13	80:20

License

This dataset is licensed under CC BY 4.0. See LICENSE.

Citation

If you use this dataset in your research, please cite the Woods Hole Oceanographic Institution and the Sosik Lab. A formal citation will be added here upon publication, including the dataset DOI and any applicable funding/award number.

Acknowledgments

Developed at the Woods Hole Oceanographic Institution. Data collected by the WHOI IFCB Lab on Northeast U.S. Shelf cruises.