Dataset Viewer
Auto-converted to Parquet Duplicate
image_id
stringlengths
9
9
filename
stringlengths
13
13
image
imagewidth (px)
512
512
category
stringclasses
10 values
split
stringclasses
1 value
pixel_size_nm
float64
0.09
6.45k
pixel_size_source
stringclasses
2 values
mag
float64
61
3.29M
βŒ€
wd_mm
float64
0
15.7
βŒ€
eht_kv
float64
0.8
25
βŒ€
detector
stringclasses
2 values
acquired_at
stringclasses
186 values
source_image_id
int64
43
126k
source_filename
stringlengths
7
35
related_doi
stringclasses
1 value
md5
stringlengths
32
32
embedding
listlengths
4.1k
4.1k
sem_00001
sem_00001.png
biological
index
37.037
computed
null
null
null
null
null
59,348
A66_8DIV_CNT_45.tif
10.23728/b2share.19cc2afd23e34b92b36a1dfd0113a89f
211d440a532bf7812dc4aaa08898bde3
[ 0.021666640415787697, 0.005286225117743015, -0.027652138844132423, -0.0037826038897037506, 0.011021725833415985, -0.009137669578194618, -0.004510864149779081, 0.00009788257739273831, -0.03402894362807274, -0.028391268104314804, -0.061159342527389526, -0.009528974071145058, -0.052724573761224...
sem_00002
sem_00002.png
biological
index
1,836
instrument
200
0
3
SE2
null
54,645
Spinals_on_MWNT_51.tif
10.23728/b2share.19cc2afd23e34b92b36a1dfd0113a89f
0a5359b0186e45c1ca9f464a686a83bf
[ 0.05190211907029152, 0.015029426664113998, -0.00995465274900198, -0.028379755094647408, -0.0031295688822865486, 0.007091191131621599, -0.035920705646276474, -0.007267346139997244, -0.025756165385246277, -0.00990967731922865, -0.06386569142341614, 0.00008175295806722715, -0.06362581998109818,...
sem_00003
sem_00003.png
biological
index
18.4
instrument
20,000
6.9
0.9
SE2
null
53,908
Cells_on_Cambridge_Graphene_09.tif
10.23728/b2share.19cc2afd23e34b92b36a1dfd0113a89f
6af5e7ed7dfcf9f7375248a834edc298
[0.01398585457354784,-0.03520526736974716,-0.008094003424048424,0.005304853431880474,-0.004867339972(...TRUNCATED)
sem_00004
sem_00004.png
biological
index
1,836
instrument
200
2.953
7
SE2
2015-08-21T10:16:29
58,450
Neurons_AGi_13072015_PdAu_13.tif
10.23728/b2share.19cc2afd23e34b92b36a1dfd0113a89f
e33daad64a518e2842071554b8f78493
[0.007108225021511316,0.0058181327767670155,-0.017123699188232422,-0.005203976295888424,0.0329100638(...TRUNCATED)
sem_00005
sem_00005.png
biological
index
4.9
instrument
75,000
7.1
2
SE2
null
59,008
SpinalsOnCNTs_23-10-12_46.tif
10.23728/b2share.19cc2afd23e34b92b36a1dfd0113a89f
336b885d88be0318d5701bba667bacd1
[0.01214005146175623,-0.002261271933093667,-0.02023341879248619,0.0012102582259103656,0.010386488400(...TRUNCATED)
sem_00006
sem_00006.png
biological
index
370.3704
computed
null
null
null
null
null
53,996
A30_8DIV_glass_03.tif
10.23728/b2share.19cc2afd23e34b92b36a1dfd0113a89f
b647dcedbfd75e85aa81367df2aa54bd
[0.005756568629294634,-0.032222162932157516,-0.00886694248765707,0.016842074692249298,-0.00325657310(...TRUNCATED)
sem_00007
sem_00007.png
biological
index
367.2
instrument
1,000
6.9
1
InLens
null
53,924
Cells_on_Cambridge_Graphene_01.tif
10.23728/b2share.19cc2afd23e34b92b36a1dfd0113a89f
9f5d4edd56b1625b9b53fd2dd9e1b369
[0.008635117672383785,0.0011415391927585006,-0.03644151613116264,-0.0011890652822330594,0.0246696770(...TRUNCATED)
sem_00008
sem_00008.png
biological
index
122.4
instrument
3,000
5.996
4
InLens
2015-04-22T09:54:05
55,986
HippoNeurons_on_GLASS09.tif
10.23728/b2share.19cc2afd23e34b92b36a1dfd0113a89f
adc1e7f639b7fad6aa693f5aa94900be
[0.014749471098184586,-0.0021182363852858543,-0.023956982418894768,0.0017554998630657792,-0.01352762(...TRUNCATED)
sem_00009
sem_00009.png
biological
index
73.5294
computed
null
null
null
null
null
59,296
A66_8DIV_GLASS_21.tif
10.23728/b2share.19cc2afd23e34b92b36a1dfd0113a89f
88cdf82b237f64bdea1ddf234c96b44b
[0.018492745235562325,0.010317322798073292,-0.01826648972928524,-0.00629371777176857,-0.002560476073(...TRUNCATED)
sem_00010
sem_00010.png
biological
index
2.0202
computed
181,470
null
null
null
null
49
im_03.tif
10.23728/b2share.19cc2afd23e34b92b36a1dfd0113a89f
ffa329328579a25a780802ea3cfb339b
[0.01836993172764778,-0.01355529110878706,-0.005732409656047821,-0.001143568311817944,0.009286940097(...TRUNCATED)
End of preview. Expand in Data Studio

SEM images for MDMC Python for Data Management II

1,234 scanning electron microscope (SEM) images in 10 categories, each with its instrument metadata, its curated category, provenance fields and a pre-computed embedding vector. It is the teaching dataset of the module Python for Data Management II of the Master in Data Management and Curation (MDMC, Area Science Park – SISSA), 2026–27.

from datasets import load_dataset

index = load_dataset("RitAreaSciencePark/sem-mdmc-2026", split="index")   # 1,184 images
query = load_dataset("RitAreaSciencePark/sem-mdmc-2026", split="query")   # 50 images
row = index[0]
row["image"]        # PIL image, 512 x 512, grayscale
row["embedding"]    # list of 4,096 floats

Source

The images come from the SEM image corpus of CNR-IOM (Trieste). The same corpus is the source of the published NFFA-EUROPE SEM dataset (Aversa et al., 2018), and this dataset uses its 10 categories. The images here are crops of the original TIFF frames at full resolution, with the instrument metadata read from TIFF tag 34118. The DOI of the published dataset is stored in every row as related_doi.

Splits

Split Rows Use
index 1,184 the images that go into the vector database
query 50 5 per category, never indexed: use them as queries and to measure retrieval
Category index query total
biological 120 5 125
fibres 113 5 118
films_coated_surface 120 5 125
mems_devices_and_electrodes 120 5 125
nanowires 115 5 120
particles 120 5 125
patterned_surface 119 5 124
porous_sponge 120 5 125
powder 120 5 125
tips 117 5 122
total 1,184 50 1,234

Fields

Field Type Layer Description
image_id string identity sem_00001 … sem_01184 (index), sem_q_01185 … sem_q_01234 (query)
filename string identity <image_id>.png
image image raw data 512 Γ— 512, 8-bit grayscale PNG: a crop of the frame at native resolution, without the instrument bar
category string curated one of the 10 categories above, assigned by scientists
split string β€” index or query
pixel_size_nm float instrument pixel size in nm
pixel_size_source string provenance instrument or computed
mag float instrument magnification
wd_mm float instrument working distance in mm
eht_kv float instrument accelerating voltage (EHT) in kV
detector string instrument InLens or SE2
acquired_at string instrument ISO date-time of acquisition
source_image_id int provenance id of the image in the CNR-IOM corpus
source_filename string provenance file name of the original TIFF
related_doi string provenance 10.23728/b2share.19cc2afd23e34b92b36a1dfd0113a89f
md5 string provenance MD5 of the PNG bytes in image
embedding list[float32], 4,096 derived vector from Qwen/Qwen3-VL-Embedding-8B

Some instrument fields are missing for some images.

Embedding provenance

embedding_model:     Qwen/Qwen3-VL-Embedding-8B
embedding_dimension: 4096
instruction:         "Represent the user's input."
input:               512 x 512 crop as PNG

Citation

If you use the images, cite the publication of the corpus they come from:

@article{aversa2018nffa,
  title   = {The first annotated set of scanning electron microscopy images for nanoscience},
  author  = {Aversa, Rossella and Modarres, Mohammad Hadi and Cozzini, Stefano and
             Ciancio, Regina and Chiusole, Alberto},
  journal = {Scientific Data},
  volume  = {5},
  pages   = {180172},
  year    = {2018},
  doi     = {10.1038/sdata.2018.172}
}

Published dataset: NFFA-EUROPE SEM dataset, DOI 10.23728/b2share.19cc2afd23e34b92b36a1dfd0113a89f.

Downloads last month
11