withoutBG Enterprise vs API Model
49 examples, 36 with ground truth. Lower is better. Use the view controls on each row to switch between transparency, chroma key, and alpha matte on all outputs.
| Metric | withoutBG Enterprise Model | withoutBG API Model |
|---|---|---|
| MGE (Edge Quality) | 0.03330 | 0.03233 ✓ |
| MAE (Overall Accuracy) | 0.03134 | 0.03071 ✓ |
| Connectivity | 0.03094 | 0.03060 ✓ |



- MGE
- 0.01013✓
- MAE
- 0.00354
- Conn.
- 0.00307
- MGE
- 0.01126
- MAE
- 0.00339✓
- Conn.
- 0.00269✓



- MGE
- 0.01897
- MAE
- 0.00695✓
- Conn.
- 0.00675
- MGE
- 0.01861✓
- MAE
- 0.00706
- Conn.
- 0.00671✓
Not Available


- MGE
- ?
- MAE
- ?
- Conn.
- ?
- MGE
- ?
- MAE
- ?
- Conn.
- ?
Photo by Alexander Dummer on Unsplash
Not Available


- MGE
- ?
- MAE
- ?
- Conn.
- ?
- MGE
- ?
- MAE
- ?
- Conn.
- ?
Not Available


- MGE
- ?
- MAE
- ?
- Conn.
- ?
- MGE
- ?
- MAE
- ?
- Conn.
- ?
Photo by Andreas Rasmussen on Unsplash



- MGE
- 0.02429✓
- MAE
- 0.00555✓
- Conn.
- 0.00509✓
- MGE
- 0.02449
- MAE
- 0.00561
- Conn.
- 0.00513



- MGE
- 0.02197✓
- MAE
- 0.02763✓
- Conn.
- 0.02760✓
- MGE
- 0.02219
- MAE
- 0.02769
- Conn.
- 0.02764



- MGE
- 0.02663
- MAE
- 0.00766
- Conn.
- 0.00765
- MGE
- 0.02617✓
- MAE
- 0.00732✓
- Conn.
- 0.00722✓



- MGE
- 0.01383
- MAE
- 0.00630
- Conn.
- 0.00614
- MGE
- 0.01370✓
- MAE
- 0.00627✓
- Conn.
- 0.00610✓



- MGE
- 0.02698
- MAE
- 0.01033✓
- Conn.
- 0.00885✓
- MGE
- 0.02480✓
- MAE
- 0.01061
- Conn.
- 0.00918
Not Available


- MGE
- ?
- MAE
- ?
- Conn.
- ?
- MGE
- ?
- MAE
- ?
- Conn.
- ?
Photo by Camila Quintero Franco on Unsplash



- MGE
- 0.01961
- MAE
- 0.01051✓
- Conn.
- 0.00992✓
- MGE
- 0.01942✓
- MAE
- 0.01082
- Conn.
- 0.01029



- MGE
- 0.05294
- MAE
- 0.01034
- Conn.
- 0.01044
- MGE
- 0.05280✓
- MAE
- 0.01033✓
- Conn.
- 0.01043✓
Not Available


- MGE
- ?
- MAE
- ?
- Conn.
- ?
- MGE
- ?
- MAE
- ?
- Conn.
- ?
Photo by Carlo Sierra on Unsplash



- MGE
- 0.01225✓
- MAE
- 0.00227✓
- Conn.
- 0.00218✓
- MGE
- 0.01234
- MAE
- 0.00296
- Conn.
- 0.00287



- MGE
- 0.02442
- MAE
- 0.00833✓
- Conn.
- 0.00834✓
- MGE
- 0.02320✓
- MAE
- 0.00840
- Conn.
- 0.00849



- MGE
- 0.02573✓
- MAE
- 0.00890
- Conn.
- 0.00852
- MGE
- 0.02596
- MAE
- 0.00885✓
- Conn.
- 0.00851✓



- MGE
- 0.02921
- MAE
- 0.00838
- Conn.
- 0.00847
- MGE
- 0.02603✓
- MAE
- 0.00683✓
- Conn.
- 0.00674✓



- MGE
- 0.00813
- MAE
- 0.00410
- Conn.
- 0.00404
- MGE
- 0.00681✓
- MAE
- 0.00344✓
- Conn.
- 0.00323✓



- MGE
- 0.01821
- MAE
- 0.00745
- Conn.
- 0.00737
- MGE
- 0.01785✓
- MAE
- 0.00705✓
- Conn.
- 0.00698✓



- MGE
- 0.03553✓
- MAE
- 0.00887✓
- Conn.
- 0.00864
- MGE
- 0.03575
- MAE
- 0.00888
- Conn.
- 0.00864✓
Not Available


- MGE
- ?
- MAE
- ?
- Conn.
- ?
- MGE
- ?
- MAE
- ?
- Conn.
- ?
Photo by Mayer Tawfik on Unsplash



- MGE
- 0.01101✓
- MAE
- 0.00580
- Conn.
- 0.00532
- MGE
- 0.01188
- MAE
- 0.00560✓
- Conn.
- 0.00515✓



- MGE
- 0.01400✓
- MAE
- 0.00546✓
- Conn.
- 0.00462✓
- MGE
- 0.01875
- MAE
- 0.00926
- Conn.
- 0.00890
Not Available


- MGE
- ?
- MAE
- ?
- Conn.
- ?
- MGE
- ?
- MAE
- ?
- Conn.
- ?
Photo by Niko Tsviliov on Unsplash



- MGE
- 0.04276✓
- MAE
- 0.02249✓
- Conn.
- 0.02242✓
- MGE
- 0.04310
- MAE
- 0.02258
- Conn.
- 0.02251
Not Available


- MGE
- ?
- MAE
- ?
- Conn.
- ?
- MGE
- ?
- MAE
- ?
- Conn.
- ?



- MGE
- 0.01986
- MAE
- 0.01650
- Conn.
- 0.01661
- MGE
- 0.01629✓
- MAE
- 0.00464✓
- Conn.
- 0.00458✓



- MGE
- 0.36815
- MAE
- 0.49492
- Conn.
- 0.49122
- MGE
- 0.33352✓
- MAE
- 0.34456✓
- Conn.
- 0.35098✓



- MGE
- 0.01895✓
- MAE
- 0.00445✓
- Conn.
- 0.00418✓
- MGE
- 0.01895
- MAE
- 0.00451
- Conn.
- 0.00426



- MGE
- 0.00678✓
- MAE
- 0.00675
- Conn.
- 0.00668
- MGE
- 0.00709
- MAE
- 0.00649✓
- Conn.
- 0.00639✓



- MGE
- 0.01831
- MAE
- 0.00536✓
- Conn.
- 0.00470✓
- MGE
- 0.01788✓
- MAE
- 0.00606
- Conn.
- 0.00521



- MGE
- 0.03321
- MAE
- 0.00544✓
- Conn.
- 0.00550
- MGE
- 0.03314✓
- MAE
- 0.00544
- Conn.
- 0.00550✓



- MGE
- 0.01406✓
- MAE
- 0.02264✓
- Conn.
- 0.02307✓
- MGE
- 0.01489
- MAE
- 0.02831
- Conn.
- 0.02966



- MGE
- 0.01292
- MAE
- 0.00283
- Conn.
- 0.00290
- MGE
- 0.01238✓
- MAE
- 0.00266✓
- Conn.
- 0.00271✓



- MGE
- 0.00968
- MAE
- 0.00244
- Conn.
- 0.00215
- MGE
- 0.00947✓
- MAE
- 0.00208✓
- Conn.
- 0.00167✓



- MGE
- 0.01032
- MAE
- 0.00463✓
- Conn.
- 0.00361✓
- MGE
- 0.01028✓
- MAE
- 0.00472
- Conn.
- 0.00378



- MGE
- 0.01947✓
- MAE
- 0.01626✓
- Conn.
- 0.01478✓
- MGE
- 0.02155
- MAE
- 0.02638
- Conn.
- 0.02497



- MGE
- ?
- MAE
- ?
- Conn.
- ?
- MGE
- ?
- MAE
- ?
- Conn.
- ?



- MGE
- 0.06552✓
- MAE
- 0.09462✓
- Conn.
- 0.09392✓
- MGE
- 0.06565
- MAE
- 0.09469
- Conn.
- 0.09398



- MGE
- 0.07561✓
- MAE
- 0.01517
- Conn.
- 0.01502
- MGE
- 0.07565
- MAE
- 0.01517✓
- Conn.
- 0.01499✓
Not Available


- MGE
- ?
- MAE
- ?
- Conn.
- ?
- MGE
- ?
- MAE
- ?
- Conn.
- ?
Photo by Vinicius Wiesehofer on Unsplash



- MGE
- 0.03824✓
- MAE
- 0.24126✓
- Conn.
- 0.24158✓
- MGE
- 0.03932
- MAE
- 0.36117
- Conn.
- 0.36178



- MGE
- 0.01538✓
- MAE
- 0.00557
- Conn.
- 0.00452
- MGE
- 0.01547
- MAE
- 0.00542✓
- Conn.
- 0.00430✓



- MGE
- 0.02655✓
- MAE
- 0.01153✓
- Conn.
- 0.01080✓
- MGE
- 0.02804
- MAE
- 0.01535
- Conn.
- 0.01514



- MGE
- 0.00930
- MAE
- 0.00714
- Conn.
- 0.00712
- MGE
- 0.00929✓
- MAE
- 0.00502✓
- Conn.
- 0.00441✓
Not Available


- MGE
- ?
- MAE
- ?
- Conn.
- ?
- MGE
- ?
- MAE
- ?
- Conn.
- ?
Photo by Conor Samuel on Unsplash
Not Available


- MGE
- ?
- MAE
- ?
- Conn.
- ?
- MGE
- ?
- MAE
- ?
- Conn.
- ?
Photo by Lars Bo Nielsen on Unsplash
Not Available


- MGE
- ?
- MAE
- ?
- Conn.
- ?
- MGE
- ?
- MAE
- ?
- Conn.
- ?
Photo by Norbert Braun on Unsplash
Methodology
- Alpha mattes are grayscale masks encoding per-pixel foreground probability. They capture fine edge detail (hair, fur, semi-transparent regions) that binary masks discard.
- Metrics are computed by comparing predicted mattes against a ground-truth reference: MGE measures edge sharpness, MAE measures pixel-level accuracy, and Connectivity penalises fragmented foreground regions. All three are lower-is-better.
- Scores are shown as ? when no ground-truth matte is available for that image. See the alpha matting evaluation benchmark for details on the evaluation methodology.
Photo Credits
Source images are courtesy of the photographers below via Unsplash.