Leaderboard - Grand Challenge

SNEMI3D: 3D Segmentation of neurites in EM images Banner

Challenge Leaderboard

Search:

#	User	Created	adapted_rand_error	Comment
1st	jasonkena	31 Oct. 2022	0.025	PNI; Kisuk Lee; PNI-UNet trained with long-range edges. Test-time augmentation (16 variants), Merge threshold = 0.665.
2nd	jasonkena	31 Oct. 2022	0.029	GAIP; Michal Januszewski; Single-pass flood-filling on 6nm data, followed by FFN agglo with alignment + orphan handling
3rd	jasonkena	11 Oct. 2022	0.035	S&T; Felix Gonda; Multi-neuron tracing with dual u-net and forward-backward consistency rev7
4th	yzyysj	1 April 2023	0.040	Transformer based model with patch size 21616
5th	jasonkena	31 Oct. 2022	0.040	CCG; Yaron Meirovitch; Cross Classification Clustering, multi-object FCN-based tracking algorithm
6th	emailofhc	14 June 2023	0.046	baseline
7th	yzyysj	12 Feb. 2023	0.046	Swin Transformer Based Model
8th	jasonkena	31 Oct. 2022	0.046	LZL-USTC; Zhili Li; RSUnet with Squeeze and excitation module, long range affinity prediction, iteration 57k
9th	jasonkena	31 Oct. 2022	0.047	CS17; Tristan Hascoet; 3rd Attempt to reproduce PNI's result, agglomeration using GALA features
10th	ivanhcenalmor	10 March 2023	0.051	Transformer-based model provided by Zhengyu Yang from Boston College (with original test images)
11th	jackie	19 June 2023	0.054	MiRA_Swin-UNETR_Full
12th	jackie	21 June 2023	0.054	MiRA_Swin-UNETR_Full_TTA_8_beta_2e-1
13th	ivanhcenalmor	13 March 2023	0.057	Transformer-based model provided by Zhengyu Yang from Boston College (test images super-resolved using RCAN model)
14th	jasonkena	31 Oct. 2022	0.058	VCG; Haidong Zhu; 3D Unet for affinity prediction, zwatershed for initial 2D segmentation, waterz for segmentation linking
15th	ivanhcenalmor	8 March 2023	0.060	Transformer-based model provided by Zhengyu Yang from Boston College (test images super-resolved using WGAN model)
16th	jasonkena	31 Oct. 2022	0.060	DIVE; Tao Zeng; Adaptive InceptionSegNet 3x3 fuse 1_3_5fm and full fuse
17th	jackie	11 June 2023	0.060	MiRA_3D_U-Net_Semi_0.08
18th	jasonkena	31 Oct. 2022	0.060	human values
19th	jasonkena	31 Oct. 2022	0.062	BCG; anon; Biologically-constrained error correction framework.
20th	ivanhcenalmor	10 March 2023	0.069	Transformer-based model provided by Zhengyu Yang from Boston College (test images super-resolved using WDSR model)
21st	ivanhcenalmor	10 March 2023	0.072	Transformer-based model provided by Zhengyu Yang from Boston College (test images super-resolved using ESRGAN+ model)
22nd	ivanhcenalmor	8 March 2023	0.073	Transformer-based model provided by Zhengyu Yang from Boston College (test images super-resolved using SR-Unet model)
23rd	yzyysj	14 Nov. 2022	0.073	U-Net with semi-supervised learning
24th	yzyysj	14 Nov. 2022	0.081	baseline obtained with transformer models (accommodated to the latest file requirements)
25th	ivanhcenalmor	8 March 2023	0.082	Transformer-based model provided by Zhengyu Yang from Boston College (test images super-resolved using DFCAN model)
26th	danifranco	3 March 2023	0.166
27th	danifranco	3 March 2023	0.167
28th	ivanhcenalmor	23 Feb. 2023	0.270	2D with 3D watershed,
29th	ivanhcenalmor	22 Feb. 2023	0.433	1000 epochs trained with Snemi data
30th	ivanhcenalmor	21 Feb. 2023	0.488	1000 epochs trained with SR-RCAN images
31st	ivanhcenalmor	19 Feb. 2023	0.801	500 epochs 0.0001 lr 128 patches - ResUnet from BiaPy BiaPy_3D_Instance_Segmentation.ipynb notebook
32nd	ivanhcenalmor	18 Feb. 2023	0.834	100 epochs 0.0001 lr - ResUnet from BiaPy BiaPy_3D_Instance_Segmentation.ipynb notebook (uint)
33rd	ivanhcenalmor	1 March 2023	0.883	Prediction - 500 epochs 160x160x18 "dense" ADAM trained with Snemi data
33rd	ivanhcenalmor	1 March 2023	0.883	Postprocessed - 500 epochs 160x160x18 "dense" ADAM trained with Snemi data
35th	ivanhcenalmor	2 March 2023	0.884	Postprocessed - simple ReducePlateau 500 epochs 160x160x18 "dense" ADAM trained with Snemi data
36th	ivanhcenalmor	22 Feb. 2023	0.889	1000 epochs 160x160x18 CONTOUR_MODE = "dense" trained with Snemi data
37th	ivanhcenalmor	22 Feb. 2023	0.893	1000 epochs 128x128x16 CONTOUR_MODE = "dense" trained with Snemi data

Showing 1 to 37 of 37 entries