Dev

Two-Photon — Dev Tier

(5 scenes)

Blind evaluation tier — no ground truth available.

Measurements (y), ideal forward operator (H), and spec ranges only.

Apply your pipeline from the Public tier. Use consistency as self-check.

Reconstructed signals and corrected spec. Scored server-side.

🔒

True spec hidden — estimate parameters from spec ranges below.

#	Method	Score	PSNR	SSIM	Consistency	Trust	Source
1	DeconvFormer + gradient	0.794	33.38	0.956	0.89	✓ Certified	Chen et al., CVPR 2024
2	Restormer+ + gradient	0.762	31.26	0.934	0.88	✓ Certified	Zamir et al., ICCV 2024
3	Restormer + gradient	0.753	31.46	0.937	0.82	✓ Certified	Zamir et al., CVPR 2022
4	DiffDeconv + gradient	0.726	29.97	0.916	0.8	✓ Certified	Huang et al., NeurIPS 2024
5	U-Net + gradient	0.722	28.54	0.892	0.9	✓ Certified	Ronneberger et al., MICCAI 2015
6	ScoreMicro + gradient	0.719	28.5	0.891	0.89	✓ Certified	Wei et al., ECCV 2025
7	TV-Deconvolution + gradient	0.680	27.02	0.859	0.83	✓ Certified	Rudin et al., Phys. A 1992
8	ResUNet + gradient	0.662	25.79	0.826	0.87	✓ Certified	DeCelle et al., Nat. Methods 2021
9	CARE + gradient	0.660	26.45	0.844	0.79	✓ Certified	Weigert et al., Nat. Methods 2018
10	Wiener Filter + gradient	0.656	25.95	0.831	0.82	✓ Certified	Analytical baseline
11	Richardson-Lucy + gradient	0.631	24.31	0.779	0.88	✓ Certified	Richardson, JOSA 1972 / Lucy, AJ 1974
12	PnP-DnCNN + gradient	0.630	24.77	0.795	0.82	✓ Certified	Zhang et al., IEEE TIP 2017
13	PnP-FISTA + gradient	0.599	23.18	0.738	0.86	✓ Certified	Bai et al., 2020

y H_ideal spec_ranges

Format: HDF5

Scenes: 5

0.4 × PSNR_norm + 0.4 × SSIM + 0.2 × (1 − ‖y − Ĥx̂‖/‖y‖)

PSNR: 40% SSIM: 40% Consistency: 20%