search

thodorii / Thesis

0 stars 0 forks source link

Shy-Morning: Baseline Unet64 S1 S2 S3 #5

Open thodorii opened 1 year ago

thodorii commented 1 year ago

Shy Morning

Train baseline model again to make sure that results match from May and nothing is broken

Hyperparameters specifications:

Filters = 64
Epochs = 30 
LR = 0.0001 (1e-4)
Scheduler = tf.keras.optimizers.schedules.ExponentialDecay(
            FLAGS.lr,
            decay_steps=200,
            decay_rate=0.96,
            staircase=True
        )

Optimizer = tf.keras.optimizers.Adam(
            learning_rate=lr_schedule,
            beta_1=0.9,
            beta_2=0.999,
            epsilon=1e-07,
            amsgrad=False,
            name='Adam',
        )
thodorii commented 1 year ago

Shy-Morning-S1

Water IoU: 58.454 Water Precision: 68.346 Water Recall: 80.154 Water F1: 73.781

592/592 [==============================] - 307s 489ms/step - loss: 38.8261 - mean_io_u: 0.3574 - val_loss: 12.7158 - val_mean_io_u: 0.4725
Epoch 2/30
592/592 [==============================] - 288s 485ms/step - loss: 8.6234 - mean_io_u: 0.4592 - val_loss: 6.4593 - val_mean_io_u: 0.5431
Epoch 3/30
592/592 [==============================] - 287s 485ms/step - loss: 5.3265 - mean_io_u: 0.5500 - val_loss: 4.5785 - val_mean_io_u: 0.6712
Epoch 4/30
592/592 [==============================] - 287s 484ms/step - loss: 4.0768 - mean_io_u: 0.6201 - val_loss: 3.6724 - val_mean_io_u: 0.7321
Epoch 5/30
592/592 [==============================] - 285s 481ms/step - loss: 3.3442 - mean_io_u: 0.7069 - val_loss: 3.1170 - val_mean_io_u: 0.7741
Epoch 6/30
592/592 [==============================] - 284s 480ms/step - loss: 2.8895 - mean_io_u: 0.7264 - val_loss: 2.7353 - val_mean_io_u: 0.7803
Epoch 7/30
592/592 [==============================] - 284s 479ms/step - loss: 2.5908 - mean_io_u: 0.7277 - val_loss: 2.4733 - val_mean_io_u: 0.7788
Epoch 8/30
592/592 [==============================] - 283s 478ms/step - loss: 2.3615 - mean_io_u: 0.7293 - val_loss: 2.2711 - val_mean_io_u: 0.7792
Epoch 9/30
592/592 [==============================] - 284s 478ms/step - loss: 2.1774 - mean_io_u: 0.7306 - val_loss: 2.1049 - val_mean_io_u: 0.7791
Epoch 10/30
592/592 [==============================] - 284s 478ms/step - loss: 2.0258 - mean_io_u: 0.7320 - val_loss: 1.9681 - val_mean_io_u: 0.7794
Epoch 11/30
592/592 [==============================] - 283s 478ms/step - loss: 1.8991 - mean_io_u: 0.7332 - val_loss: 1.8503 - val_mean_io_u: 0.7787
Epoch 12/30
592/592 [==============================] - 283s 478ms/step - loss: 1.7922 - mean_io_u: 0.7344 - val_loss: 1.7494 - val_mean_io_u: 0.7773
Epoch 13/30
592/592 [==============================] - 283s 478ms/step - loss: 1.7012 - mean_io_u: 0.7355 - val_loss: 1.6625 - val_mean_io_u: 0.7748
Epoch 14/30
592/592 [==============================] - 283s 478ms/step - loss: 1.6234 - mean_io_u: 0.7363 - val_loss: 1.5878 - val_mean_io_u: 0.7708
Epoch 15/30
592/592 [==============================] - 283s 477ms/step - loss: 1.5568 - mean_io_u: 0.7370 - val_loss: 1.5246 - val_mean_io_u: 0.7658
Epoch 16/30
592/592 [==============================] - 283s 477ms/step - loss: 1.4995 - mean_io_u: 0.7377 - val_loss: 1.4713 - val_mean_io_u: 0.7603
Epoch 17/30
592/592 [==============================] - 283s 477ms/step - loss: 1.4501 - mean_io_u: 0.7382 - val_loss: 1.4260 - val_mean_io_u: 0.7563
Epoch 18/30
592/592 [==============================] - 283s 477ms/step - loss: 1.4077 - mean_io_u: 0.7387 - val_loss: 1.3873 - val_mean_io_u: 0.7542
Epoch 19/30
592/592 [==============================] - 282s 477ms/step - loss: 1.3711 - mean_io_u: 0.7391 - val_loss: 1.3538 - val_mean_io_u: 0.7540
Epoch 20/30
592/592 [==============================] - 282s 477ms/step - loss: 1.3397 - mean_io_u: 0.7395 - val_loss: 1.3250 - val_mean_io_u: 0.7550
Epoch 21/30
592/592 [==============================] - 284s 478ms/step - loss: 1.3127 - mean_io_u: 0.7400 - val_loss: 1.3001 - val_mean_io_u: 0.7570
Epoch 22/30
592/592 [==============================] - 283s 478ms/step - loss: 1.2897 - mean_io_u: 0.7404 - val_loss: 1.2789 - val_mean_io_u: 0.7590
Epoch 23/30
592/592 [==============================] - 283s 476ms/step - loss: 1.2700 - mean_io_u: 0.7408 - val_loss: 1.2609 - val_mean_io_u: 0.7608
Epoch 24/30
592/592 [==============================] - 283s 477ms/step - loss: 1.2532 - mean_io_u: 0.7411 - val_loss: 1.2455 - val_mean_io_u: 0.7619
Epoch 25/30
592/592 [==============================] - 283s 477ms/step - loss: 1.2388 - mean_io_u: 0.7414 - val_loss: 1.2324 - val_mean_io_u: 0.7629
Epoch 26/30
592/592 [==============================] - 283s 477ms/step - loss: 1.2265 - mean_io_u: 0.7417 - val_loss: 1.2211 - val_mean_io_u: 0.7633
Epoch 27/30
592/592 [==============================] - 283s 477ms/step - loss: 1.2160 - mean_io_u: 0.7419 - val_loss: 1.2114 - val_mean_io_u: 0.7632
Epoch 28/30
592/592 [==============================] - 283s 477ms/step - loss: 1.2070 - mean_io_u: 0.7421 - val_loss: 1.2031 - val_mean_io_u: 0.7627
Epoch 29/30
592/592 [==============================] - 283s 477ms/step - loss: 1.1993 - mean_io_u: 0.7423 - val_loss: 1.1959 - val_mean_io_u: 0.7618
Epoch 30/30
592/592 [==============================] - 283s 478ms/step - loss: 1.1926 - mean_io_u: 0.7425 - val_loss: 1.1897 - val_mean_io_u: 0.7608
thodorii commented 1 year ago

Shy-Morning-S2

Water IoU: 58.685 Water Precision: 69.474 Water Recall: 79.073 Water F1: 73.964

592/592 [==============================] - 316s 503ms/step - loss: 36.6388 - mean_io_u: 0.4735 - val_loss: 10.8529 - val_mean_io_u: 0.6049
Epoch 2/30
592/592 [==============================] - 295s 498ms/step - loss: 7.3874 - mean_io_u: 0.5768 - val_loss: 5.3856 - val_mean_io_u: 0.6544
Epoch 3/30
592/592 [==============================] - 299s 505ms/step - loss: 4.4891 - mean_io_u: 0.6230 - val_loss: 3.8046 - val_mean_io_u: 0.7042
Epoch 4/30
592/592 [==============================] - 300s 506ms/step - loss: 3.3590 - mean_io_u: 0.6961 - val_loss: 3.0231 - val_mean_io_u: 0.7644
Epoch 5/30
592/592 [==============================] - 302s 509ms/step - loss: 2.7492 - mean_io_u: 0.7285 - val_loss: 2.5236 - val_mean_io_u: 0.7665
Epoch 6/30
592/592 [==============================] - 298s 502ms/step - loss: 2.3581 - mean_io_u: 0.7332 - val_loss: 2.1901 - val_mean_io_u: 0.7594
Epoch 7/30
592/592 [==============================] - 302s 511ms/step - loss: 2.0747 - mean_io_u: 0.7358 - val_loss: 1.9444 - val_mean_io_u: 0.7518
Epoch 8/30
592/592 [==============================] - 302s 511ms/step - loss: 1.8589 - mean_io_u: 0.7377 - val_loss: 1.7566 - val_mean_io_u: 0.7419
Epoch 9/30
592/592 [==============================] - 298s 502ms/step - loss: 1.6900 - mean_io_u: 0.7394 - val_loss: 1.6092 - val_mean_io_u: 0.7363
Epoch 10/30
592/592 [==============================] - 301s 508ms/step - loss: 1.5558 - mean_io_u: 0.7408 - val_loss: 1.4912 - val_mean_io_u: 0.7331
Epoch 11/30
592/592 [==============================] - 303s 510ms/step - loss: 1.4478 - mean_io_u: 0.7419 - val_loss: 1.3959 - val_mean_io_u: 0.7314
Epoch 12/30
592/592 [==============================] - 298s 502ms/step - loss: 1.3602 - mean_io_u: 0.7429 - val_loss: 1.3181 - val_mean_io_u: 0.7308
Epoch 13/30
592/592 [==============================] - 298s 503ms/step - loss: 1.2889 - mean_io_u: 0.7437 - val_loss: 1.2543 - val_mean_io_u: 0.7314
Epoch 14/30
592/592 [==============================] - 300s 506ms/step - loss: 1.2304 - mean_io_u: 0.7445 - val_loss: 1.2017 - val_mean_io_u: 0.7328
Epoch 15/30
592/592 [==============================] - 299s 504ms/step - loss: 1.1824 - mean_io_u: 0.7451 - val_loss: 1.1584 - val_mean_io_u: 0.7343
Epoch 16/30
592/592 [==============================] - 299s 504ms/step - loss: 1.1428 - mean_io_u: 0.7457 - val_loss: 1.1226 - val_mean_io_u: 0.7357
Epoch 17/30
592/592 [==============================] - 297s 502ms/step - loss: 1.1102 - mean_io_u: 0.7462 - val_loss: 1.0930 - val_mean_io_u: 0.7373
Epoch 18/30
592/592 [==============================] - 297s 502ms/step - loss: 1.0832 - mean_io_u: 0.7466 - val_loss: 1.0683 - val_mean_io_u: 0.7393
Epoch 19/30
592/592 [==============================] - 296s 499ms/step - loss: 1.0607 - mean_io_u: 0.7470 - val_loss: 1.0477 - val_mean_io_u: 0.7418
Epoch 20/30
592/592 [==============================] - 298s 502ms/step - loss: 1.0420 - mean_io_u: 0.7473 - val_loss: 1.0305 - val_mean_io_u: 0.7444
Epoch 21/30
592/592 [==============================] - 293s 495ms/step - loss: 1.0264 - mean_io_u: 0.7476 - val_loss: 1.0160 - val_mean_io_u: 0.7470
Epoch 22/30
592/592 [==============================] - 293s 495ms/step - loss: 1.0132 - mean_io_u: 0.7478 - val_loss: 1.0039 - val_mean_io_u: 0.7495
Epoch 23/30
592/592 [==============================] - 298s 502ms/step - loss: 1.0020 - mean_io_u: 0.7480 - val_loss: 0.9938 - val_mean_io_u: 0.7517
Epoch 24/30
592/592 [==============================] - 295s 497ms/step - loss: 0.9926 - mean_io_u: 0.7482 - val_loss: 0.9852 - val_mean_io_u: 0.7535
Epoch 25/30
592/592 [==============================] - 296s 499ms/step - loss: 0.9845 - mean_io_u: 0.7483 - val_loss: 0.9779 - val_mean_io_u: 0.7545
Epoch 26/30
592/592 [==============================] - 294s 493ms/step - loss: 0.9775 - mean_io_u: 0.7485 - val_loss: 0.9717 - val_mean_io_u: 0.7554
Epoch 27/30
592/592 [==============================] - 297s 501ms/step - loss: 0.9715 - mean_io_u: 0.7486 - val_loss: 0.9664 - val_mean_io_u: 0.7559
Epoch 28/30
592/592 [==============================] - 295s 497ms/step - loss: 0.9664 - mean_io_u: 0.7486 - val_loss: 0.9617 - val_mean_io_u: 0.7562
Epoch 29/30
592/592 [==============================] - 295s 498ms/step - loss: 0.9619 - mean_io_u: 0.7487 - val_loss: 0.9577 - val_mean_io_u: 0.7564
Epoch 30/30
592/592 [==============================] - 294s 496ms/step - loss: 0.9581 - mean_io_u: 0.7488 - val_loss: 0.9542 - val_mean_io_u: 0.7566
thodorii commented 1 year ago

Shy-Morning S3

Water IoU: 60.063 Water Precision: 70.926 Water Recall: 79.681 Water F1: 75.049

592/592 [==============================] - 335s 534ms/step - loss: 36.3658 - mean_io_u: 0.4981 - val_loss: 10.3519 - val_mean_io_u: 0.6458
Epoch 2/30
592/592 [==============================] - 327s 551ms/step - loss: 6.9122 - mean_io_u: 0.6168 - val_loss: 4.9705 - val_mean_io_u: 0.7171
Epoch 3/30
592/592 [==============================] - 340s 574ms/step - loss: 4.1454 - mean_io_u: 0.6763 - val_loss: 3.5238 - val_mean_io_u: 0.7507
Epoch 4/30
592/592 [==============================] - 338s 570ms/step - loss: 3.1421 - mean_io_u: 0.7284 - val_loss: 2.8709 - val_mean_io_u: 0.7701
Epoch 5/30
592/592 [==============================] - 322s 544ms/step - loss: 2.6016 - mean_io_u: 0.7367 - val_loss: 2.4635 - val_mean_io_u: 0.7715
Epoch 6/30
592/592 [==============================] - 322s 543ms/step - loss: 2.2449 - mean_io_u: 0.7394 - val_loss: 2.1735 - val_mean_io_u: 0.7711
Epoch 7/30
592/592 [==============================] - 326s 549ms/step - loss: 1.9855 - mean_io_u: 0.7411 - val_loss: 1.9428 - val_mean_io_u: 0.7729
Epoch 8/30
592/592 [==============================] - 320s 539ms/step - loss: 1.7883 - mean_io_u: 0.7423 - val_loss: 1.7620 - val_mean_io_u: 0.7747
Epoch 9/30
592/592 [==============================] - 321s 541ms/step - loss: 1.6350 - mean_io_u: 0.7431 - val_loss: 1.6146 - val_mean_io_u: 0.7761
Epoch 10/30
592/592 [==============================] - 319s 536ms/step - loss: 1.5139 - mean_io_u: 0.7439 - val_loss: 1.4934 - val_mean_io_u: 0.7761
Epoch 11/30
592/592 [==============================] - 312s 526ms/step - loss: 1.4173 - mean_io_u: 0.7446 - val_loss: 1.3961 - val_mean_io_u: 0.7737
Epoch 12/30
592/592 [==============================] - 312s 526ms/step - loss: 1.3398 - mean_io_u: 0.7453 - val_loss: 1.3211 - val_mean_io_u: 0.7709
Epoch 13/30
592/592 [==============================] - 311s 525ms/step - loss: 1.2770 - mean_io_u: 0.7461 - val_loss: 1.2634 - val_mean_io_u: 0.7693
Epoch 14/30
592/592 [==============================] - 311s 523ms/step - loss: 1.2261 - mean_io_u: 0.7468 - val_loss: 1.2188 - val_mean_io_u: 0.7698
Epoch 15/30
592/592 [==============================] - 316s 533ms/step - loss: 1.1847 - mean_io_u: 0.7474 - val_loss: 1.1835 - val_mean_io_u: 0.7708
Epoch 16/30
592/592 [==============================] - 314s 528ms/step - loss: 1.1508 - mean_io_u: 0.7480 - val_loss: 1.1546 - val_mean_io_u: 0.7717
Epoch 17/30
592/592 [==============================] - 312s 526ms/step - loss: 1.1230 - mean_io_u: 0.7485 - val_loss: 1.1302 - val_mean_io_u: 0.7721
Epoch 18/30
592/592 [==============================] - 312s 525ms/step - loss: 1.1001 - mean_io_u: 0.7490 - val_loss: 1.1090 - val_mean_io_u: 0.7720
Epoch 19/30
592/592 [==============================] - 313s 528ms/step - loss: 1.0812 - mean_io_u: 0.7494 - val_loss: 1.0906 - val_mean_io_u: 0.7713
Epoch 20/30
592/592 [==============================] - 322s 542ms/step - loss: 1.0656 - mean_io_u: 0.7497 - val_loss: 1.0749 - val_mean_io_u: 0.7704
Epoch 21/30
592/592 [==============================] - 331s 558ms/step - loss: 1.0525 - mean_io_u: 0.7500 - val_loss: 1.0616 - val_mean_io_u: 0.7692
Epoch 22/30
592/592 [==============================] - 323s 545ms/step - loss: 1.0416 - mean_io_u: 0.7502 - val_loss: 1.0502 - val_mean_io_u: 0.7677
Epoch 23/30
592/592 [==============================] - 324s 546ms/step - loss: 1.0324 - mean_io_u: 0.7504 - val_loss: 1.0407 - val_mean_io_u: 0.7663
Epoch 24/30
592/592 [==============================] - 324s 546ms/step - loss: 1.0246 - mean_io_u: 0.7505 - val_loss: 1.0327 - val_mean_io_u: 0.7648
Epoch 25/30
592/592 [==============================] - 324s 542ms/step - loss: 1.0180 - mean_io_u: 0.7506 - val_loss: 1.0263 - val_mean_io_u: 0.7642
Epoch 26/30
592/592 [==============================] - 325s 549ms/step - loss: 1.0123 - mean_io_u: 0.7507 - val_loss: 1.0209 - val_mean_io_u: 0.7637
Epoch 27/30
592/592 [==============================] - 341s 574ms/step - loss: 1.0074 - mean_io_u: 0.7508 - val_loss: 1.0164 - val_mean_io_u: 0.7636
Epoch 28/30
592/592 [==============================] - 325s 546ms/step - loss: 1.0032 - mean_io_u: 0.7509 - val_loss: 1.0127 - val_mean_io_u: 0.7638
Epoch 29/30
592/592 [==============================] - 330s 557ms/step - loss: 0.9995 - mean_io_u: 0.7510 - val_loss: 1.0095 - val_mean_io_u: 0.7640
Epoch 30/30
592/592 [==============================] - 334s 562ms/step - loss: 0.9963 - mean_io_u: 0.7512 - val_loss: 1.0067 - val_mean_io_u: 0.7640