Link paper and GitHub repository, add task category

#2
by nielsr HF Staff - opened
Files changed (1) hide show
  1. README.md +23 -17
README.md CHANGED
@@ -1,28 +1,34 @@
1
  ---
2
  license: other
 
 
3
  pretty_name: ConnectomeBench2
 
 
4
  tags:
5
- - connectomics
6
- - proofreading
7
- - 3d
8
- - electron-microscopy
9
- - mesh
10
- size_categories:
11
- - 100K<n<1M
12
  configs:
13
- - config_name: default
14
- data_files:
15
- - split: train
16
- path: train/train-*.parquet
17
- - split: validation
18
- path: val/val-*.parquet
19
- - split: test
20
- path: test/test-*.parquet
21
  ---
22
 
23
  # ConnectomeBench2
24
 
25
- ConnectomeBench2 is a unified benchmark for **automated proofreading of connectomic neural-segmentation data**. **401,170 samples** across 4 species (mouse, fly, human, zebrafish) and 5 sample types (real merge edits, real split edits, synthetic adjacent / junction / synapse controls), with the associated mesh geometry and electron-microscopy (EM) renderings.
 
 
 
 
26
 
27
  Downstream trainers should treat this dataset as the single source of truth for sample identity, labels, train/validation/test split, and which task(s) a row is valid for.
28
 
@@ -205,4 +211,4 @@ Please also cite the upstream connectome sources used by this dataset:
205
  - MICrONS (mouse cortex): https://www.microns-explorer.org/cortical-mm3
206
  - FlyWire (Drosophila): https://flywire.ai/
207
  - H01 (human cortex): https://h01-release.storage.googleapis.com/landing.html
208
- - Zebrafish (fish1): https://fish1-release.storage.googleapis.com/index.html
 
1
  ---
2
  license: other
3
+ size_categories:
4
+ - 100K<n<1M
5
  pretty_name: ConnectomeBench2
6
+ task_categories:
7
+ - image-classification
8
  tags:
9
+ - connectomics
10
+ - proofreading
11
+ - 3d
12
+ - electron-microscopy
13
+ - mesh
 
 
14
  configs:
15
+ - config_name: default
16
+ data_files:
17
+ - split: train
18
+ path: train/train-*.parquet
19
+ - split: validation
20
+ path: val/val-*.parquet
21
+ - split: test
22
+ path: test/test-*.parquet
23
  ---
24
 
25
  # ConnectomeBench2
26
 
27
+ ConnectomeBench2 is a unified benchmark for **automated proofreading of connectomic neural-segmentation data**. This is the dataset for the paper [ConnectomeBench2: A Unified Benchmark for Automated Connectomic Proofreading](https://huggingface.co/papers/2606.21116).
28
+
29
+ The official codebase is available on GitHub at [timfarkas/ConnectomeBench2](https://github.com/timfarkas/ConnectomeBench2).
30
+
31
+ **401,170 samples** across 4 species (mouse, fly, human, zebrafish) and 5 sample types (real merge edits, real split edits, synthetic adjacent / junction / synapse controls), with the associated mesh geometry and electron-microscopy (EM) renderings.
32
 
33
  Downstream trainers should treat this dataset as the single source of truth for sample identity, labels, train/validation/test split, and which task(s) a row is valid for.
34
 
 
211
  - MICrONS (mouse cortex): https://www.microns-explorer.org/cortical-mm3
212
  - FlyWire (Drosophila): https://flywire.ai/
213
  - H01 (human cortex): https://h01-release.storage.googleapis.com/landing.html
214
+ - Zebrafish (fish1): https://fish1-release.storage.googleapis.com/index.html