swin-tiny-patch4-window7-224-finetuned-vit

This model is a fine-tuned version of microsoft/swin-tiny-patch4-window7-224 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.5493
• Crack: {'precision': 0.5735294117647058, 'recall': 0.6964285714285714, 'f1-score': 0.6290322580645161, 'support': 56}
• Environment - ground: {'precision': 0.9722222222222222, 'recall': 0.9722222222222222, 'f1-score': 0.9722222222222222, 'support': 36}
• Environment - other: {'precision': 0.8913043478260869, 'recall': 0.9318181818181818, 'f1-score': 0.9111111111111111, 'support': 44}
• Environment - sky: {'precision': 0.8974358974358975, 'recall': 0.9722222222222222, 'f1-score': 0.9333333333333333, 'support': 36}
• Environment - vegetation: {'precision': 0.9622641509433962, 'recall': 0.9622641509433962, 'f1-score': 0.9622641509433962, 'support': 53}
• Joint defect: {'precision': 0.6470588235294118, 'recall': 0.7857142857142857, 'f1-score': 0.7096774193548386, 'support': 28}
• Loss of section: {'precision': 0.0, 'recall': 0.0, 'f1-score': 0.0, 'support': 3}
• Spalling: {'precision': 0.5161290322580645, 'recall': 0.41025641025641024, 'f1-score': 0.4571428571428572, 'support': 39}
• Vegetation: {'precision': 0.8387096774193549, 'recall': 0.8813559322033898, 'f1-score': 0.859504132231405, 'support': 59}
• Wall - grafitti: {'precision': 0.96, 'recall': 0.96, 'f1-score': 0.96, 'support': 25}
• Wall - normal: {'precision': 0.7575757575757576, 'recall': 0.625, 'f1-score': 0.6849315068493151, 'support': 40}
• Wall - other: {'precision': 0.8888888888888888, 'recall': 0.8333333333333334, 'f1-score': 0.8602150537634408, 'support': 48}
• Wall - stain: {'precision': 0.84, 'recall': 0.7636363636363637, 'f1-score': 0.8000000000000002, 'support': 55}
• Accuracy: 0.8084
• Macro avg: {'precision': 0.7496244776818297, 'recall': 0.753403974906029, 'f1-score': 0.7491872342320336, 'support': 522}
• Weighted avg: {'precision': 0.805637888745576, 'recall': 0.8084291187739464, 'f1-score': 0.8046217646882747, 'support': 522}

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Crack	Environment - ground	Environment - other	Environment - sky	Environment - vegetation	Joint defect	Loss of section	Spalling	Vegetation	Wall - grafitti	Wall - normal	Wall - other	Wall - stain	Accuracy	Macro avg	Weighted avg
0.8877	0.99	146	0.8076	{'precision': 0.5294117647058824, 'recall': 0.6428571428571429, 'f1-score': 0.5806451612903226, 'support': 56}	{'precision': 0.9459459459459459, 'recall': 0.9722222222222222, 'f1-score': 0.9589041095890412, 'support': 36}	{'precision': 0.7916666666666666, 'recall': 0.8636363636363636, 'f1-score': 0.8260869565217391, 'support': 44}	{'precision': 0.8780487804878049, 'recall': 1.0, 'f1-score': 0.9350649350649352, 'support': 36}	{'precision': 0.9807692307692307, 'recall': 0.9622641509433962, 'f1-score': 0.9714285714285713, 'support': 53}	{'precision': 0.7037037037037037, 'recall': 0.6785714285714286, 'f1-score': 0.6909090909090909, 'support': 28}	{'precision': 0.0, 'recall': 0.0, 'f1-score': 0.0, 'support': 3}	{'precision': 0.5588235294117647, 'recall': 0.48717948717948717, 'f1-score': 0.5205479452054794, 'support': 39}	{'precision': 0.6625, 'recall': 0.8983050847457628, 'f1-score': 0.762589928057554, 'support': 59}	{'precision': 0.7666666666666667, 'recall': 0.92, 'f1-score': 0.8363636363636363, 'support': 25}	{'precision': 0.9411764705882353, 'recall': 0.4, 'f1-score': 0.5614035087719298, 'support': 40}	{'precision': 0.8780487804878049, 'recall': 0.75, 'f1-score': 0.8089887640449439, 'support': 48}	{'precision': 0.7659574468085106, 'recall': 0.6545454545454545, 'f1-score': 0.7058823529411765, 'support': 55}	0.7625	{'precision': 0.7232860758647859, 'recall': 0.7099677949770199, 'f1-score': 0.7045242277068015, 'support': 522}	{'precision': 0.7735594241725829, 'recall': 0.7624521072796935, 'f1-score': 0.7551578669008161, 'support': 522}
0.8113	2.0	293	0.6101	{'precision': 0.5555555555555556, 'recall': 0.7142857142857143, 'f1-score': 0.6250000000000001, 'support': 56}	{'precision': 0.9714285714285714, 'recall': 0.9444444444444444, 'f1-score': 0.9577464788732395, 'support': 36}	{'precision': 0.8888888888888888, 'recall': 0.9090909090909091, 'f1-score': 0.8988764044943819, 'support': 44}	{'precision': 0.8974358974358975, 'recall': 0.9722222222222222, 'f1-score': 0.9333333333333333, 'support': 36}	{'precision': 0.9622641509433962, 'recall': 0.9622641509433962, 'f1-score': 0.9622641509433962, 'support': 53}	{'precision': 0.7857142857142857, 'recall': 0.7857142857142857, 'f1-score': 0.7857142857142857, 'support': 28}	{'precision': 0.0, 'recall': 0.0, 'f1-score': 0.0, 'support': 3}	{'precision': 0.4117647058823529, 'recall': 0.5384615384615384, 'f1-score': 0.4666666666666667, 'support': 39}	{'precision': 0.8153846153846154, 'recall': 0.8983050847457628, 'f1-score': 0.8548387096774194, 'support': 59}	{'precision': 0.7741935483870968, 'recall': 0.96, 'f1-score': 0.8571428571428571, 'support': 25}	{'precision': 0.75, 'recall': 0.45, 'f1-score': 0.5625000000000001, 'support': 40}	{'precision': 0.9024390243902439, 'recall': 0.7708333333333334, 'f1-score': 0.8314606741573034, 'support': 48}	{'precision': 0.8157894736842105, 'recall': 0.5636363636363636, 'f1-score': 0.6666666666666666, 'support': 55}	0.7778	{'precision': 0.7331429782842396, 'recall': 0.7284044651444593, 'f1-score': 0.7232469405899653, 'support': 522}	{'precision': 0.7896708656541912, 'recall': 0.7777777777777778, 'f1-score': 0.7754219719596664, 'support': 522}
0.6069	2.98	438	0.5493	{'precision': 0.5735294117647058, 'recall': 0.6964285714285714, 'f1-score': 0.6290322580645161, 'support': 56}	{'precision': 0.9722222222222222, 'recall': 0.9722222222222222, 'f1-score': 0.9722222222222222, 'support': 36}	{'precision': 0.8913043478260869, 'recall': 0.9318181818181818, 'f1-score': 0.9111111111111111, 'support': 44}	{'precision': 0.8974358974358975, 'recall': 0.9722222222222222, 'f1-score': 0.9333333333333333, 'support': 36}	{'precision': 0.9622641509433962, 'recall': 0.9622641509433962, 'f1-score': 0.9622641509433962, 'support': 53}	{'precision': 0.6470588235294118, 'recall': 0.7857142857142857, 'f1-score': 0.7096774193548386, 'support': 28}	{'precision': 0.0, 'recall': 0.0, 'f1-score': 0.0, 'support': 3}	{'precision': 0.5161290322580645, 'recall': 0.41025641025641024, 'f1-score': 0.4571428571428572, 'support': 39}	{'precision': 0.8387096774193549, 'recall': 0.8813559322033898, 'f1-score': 0.859504132231405, 'support': 59}	{'precision': 0.96, 'recall': 0.96, 'f1-score': 0.96, 'support': 25}	{'precision': 0.7575757575757576, 'recall': 0.625, 'f1-score': 0.6849315068493151, 'support': 40}	{'precision': 0.8888888888888888, 'recall': 0.8333333333333334, 'f1-score': 0.8602150537634408, 'support': 48}	{'precision': 0.84, 'recall': 0.7636363636363637, 'f1-score': 0.8000000000000002, 'support': 55}	0.8084	{'precision': 0.7496244776818297, 'recall': 0.753403974906029, 'f1-score': 0.7491872342320336, 'support': 522}	{'precision': 0.805637888745576, 'recall': 0.8084291187739464, 'f1-score': 0.8046217646882747, 'support': 522}

Framework versions

• Transformers 4.33.1
• Pytorch 2.0.1+cu118
• Datasets 2.14.5
• Tokenizers 0.13.3