Image Classification
Keras
LiteRT
TF-Keras
Safetensors
English
efficientnetv2-s
efficientnetv2
fgic
transfer-learning
gem-pooling
focal-loss
swa
grad-cam
calibration
temperature-scaling
computer-vision
tensorflow.js
Eval Results (legacy)
Instructions to use 0xgr3y/Arch-Building-Image-Classification with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Keras
How to use 0xgr3y/Arch-Building-Image-Classification with Keras:
# Available backend options are: "jax", "torch", "tensorflow". import os os.environ["KERAS_BACKEND"] = "jax" import keras model = keras.saving.load_model("hf://0xgr3y/Arch-Building-Image-Classification") - Notebooks
- Google Colab
- Kaggle
File size: 3,862 Bytes
202ce04 83bee53 202ce04 0fe9a57 202ce04 0fe9a57 202ce04 0fe9a57 202ce04 0fe9a57 202ce04 735a201 0fe9a57 735a201 0fe9a57 735a201 0fe9a57 735a201 0fe9a57 202ce04 735a201 0fe9a57 202ce04 735a201 202ce04 735a201 202ce04 735a201 202ce04 735a201 0fe9a57 735a201 0fe9a57 202ce04 0fe9a57 202ce04 735a201 202ce04 0fe9a57 735a201 0fe9a57 735a201 202ce04 735a201 0fe9a57 735a201 202ce04 0fe9a57 735a201 202ce04 735a201 0fe9a57 735a201 202ce04 735a201 0fe9a57 735a201 0fe9a57 735a201 0fe9a57 202ce04 0fe9a57 5de52cc 735a201 202ce04 9a41f78 5de52cc 0fe9a57 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 | {
"model_type": "efficientnetv2-s",
"base_model": "efficientnetv2-s",
"task": "image-classification",
"num_labels": 8,
"id2label": {
"0": "barn",
"1": "bridge",
"2": "castle",
"3": "mosque",
"4": "skyscraper",
"5": "stadium",
"6": "temple",
"7": "windmill"
},
"label2id": {
"barn": 0,
"bridge": 1,
"castle": 2,
"mosque": 3,
"skyscraper": 4,
"stadium": 5,
"temple": 6,
"windmill": 7
},
"input_shape": [
320,
320,
3
],
"preprocessing": {
"mode": "efficientnet_v2_preprocess_input",
"mean": [
0.0,
0.0,
0.0
],
"std": [
1.0,
1.0,
1.0
],
"note": "preprocess_input is identity in TF 2.12+; EfficientNetV2-S includes internal Rescaling layer. Input expects raw [0, 255] float32.",
"channel_order": "RGB"
},
"training_info": {
"dataset_source": "Pexels Architectural Buildings (13440 images, 8 classes, balanced)",
"dataset_size": 13440,
"split_ratio": [
0.8,
0.1,
0.1
],
"split_seed": 42,
"phases": [
{
"name": "Fase 1 - Feature Extraction (Head)",
"optimizer": "AdamW",
"lr": 0.001,
"epochs_max": 25,
"epochs_actual": 1,
"val_accuracy": 0.9234,
"val_loss": 1.0109,
"train_accuracy": 0.5696,
"cutmix_mixup": true,
"label_smoothing": 0.1,
"early_stop_reason": "myCallback val_acc >= 0.85"
},
{
"name": "Fase 2 - Selective Fine-Tuning",
"optimizer": "DiscriminativeAdamW",
"lr": 0.0003,
"epochs_max": 50,
"epochs_actual": 1,
"val_accuracy": 0.9628,
"val_loss": 0.5655,
"train_accuracy": 0.8496,
"unfreeze": "block6+top_conv (BN frozen)",
"discriminative_lr": {
"block6": 0.1
},
"cutmix_mixup": false,
"label_smoothing": 0.05,
"early_stop_reason": "myCallback val_acc >= 0.92"
},
{
"name": "SWA Post-Training",
"epochs": 10,
"lr": 0.0001,
"bn_update": true,
"bn_update_steps": 100,
"val_accuracy": 0.9836,
"val_loss": 0.4109,
"method": "Izmailov et al., UAI 2018"
}
],
"metrics": {
"train_accuracy": 0.9988,
"val_accuracy": 0.9836,
"test_accuracy": 0.9777,
"test_loss": 0.4262,
"tta_accuracy": 0.9799,
"overfitting_gap": 0.0211,
"test_correct": 1314,
"test_total": 1344,
"macro_precision": 0.9777,
"macro_recall": 0.9777,
"macro_f1": 0.9777,
"per_class_f1": {
"barn": 0.9731,
"bridge": 0.9676,
"castle": 0.9792,
"mosque": 0.9792,
"skyscraper": 0.994,
"stadium": 0.9791,
"temple": 0.9668,
"windmill": 0.9822
},
"per_class_recall": {
"barn": 0.9702,
"bridge": 0.9762,
"castle": 0.9821,
"mosque": 0.9821,
"skyscraper": 0.994,
"stadium": 0.9762,
"temple": 0.9524,
"windmill": 0.9881
},
"checkpoint_comparison": {
"fine_tuning_swa": {
"val_accuracy": 0.9836,
"val_loss": 0.4109,
"rank": 0
},
"fine_tuning": {
"val_accuracy": 0.9628,
"val_loss": 0.5655,
"rank": 1
},
"fine_tuning_ema": {
"val_accuracy": 0.9353,
"val_loss": 0.6007,
"rank": 2
},
"head_training": {
"val_accuracy": 0.9234,
"val_loss": 1.0109,
"rank": 3
}
}
}
},
"version": "v6",
"license": "apache-2.0",
"github": "https://github.com/arcxteam/building-architectural-image-classifier",
"author": {
"name": "Saugani",
"email": "team@greyscope.xyz"
}
} |