Upload EuroSAT classifier

.gitattributes

@@ -33,3 +33,9 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+confusion_matrix_detailed.png filter=lfs diff=lfs merge=lfs -text
+confusion_matrix_epoch_1.png filter=lfs diff=lfs merge=lfs -text
+confusion_matrix_epoch_6.png filter=lfs diff=lfs merge=lfs -text
+per_class_performance.png filter=lfs diff=lfs merge=lfs -text
+probability_distribution.png filter=lfs diff=lfs merge=lfs -text
+training_history.png filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,99 @@
+---
+        license: apache-2.0
+        base_model: microsoft/swin-base-patch4-window7-224
+        tags:
+        - image-classification
+        - satellite-imagery
+        - eurosat
+        - remote-sensing
+        - transformer
+        datasets:
+        - nielsr/eurosat-demo
+        metrics:
+        - accuracy
+        - f1
+        library_name: transformers
+        pipeline_tag: image-classification
+        ---
+        
+        # EuroSAT Satellite Image Classifier
+        
+        This model is a fine-tuned version of `microsoft/swin-base-patch4-window7-224` for satellite image classification on the EuroSAT-SAR dataset.
+        
+        ## Model Details
+        
+        - **Base Model**: microsoft/swin-base-patch4-window7-224
+        - **Dataset**: EuroSAT-SAR (Synthetic Aperture Radar satellite images)
+        - **Task**: Multi-class image classification
+        - **Number of Classes**: 10
+        - **Image Size**: 224x224
+        
+        ## Classes
+        
+        The model can classify satellite images into the following 10 categories:
+        
+        - 0: AnnualCrop
+- 1: Forest
+- 2: HerbaceousVegetation
+- 3: Highway
+- 4: Industrial
+- 5: Pasture
+- 6: PermanentCrop
+- 7: Residential
+- 8: River
+- 9: SeaLake
+        
+        ## Training Details
+        
+        - **Learning Rate**: 5e-05
+        - **Batch Size**: 32
+        - **Number of Epochs**: 10
+        - **Optimizer**: AdamW
+        - **Weight Decay**: 0.01
+        - **Warmup Steps**: 500
+        
+        ## Usage
+        
+        ```python
+        from transformers import AutoImageProcessor, AutoModel
+        from PIL import Image
+        import torch
+        
+        # Load the model and processor
+        processor = AutoImageProcessor.from_pretrained("Adilbai/eurosat-swin-transformer")
+        model = AutoModel.from_pretrained("Adilbai/eurosat-swin-transformer")
+        
+        # Load and preprocess an image
+        image = Image.open("path_to_your_satellite_image.jpg")
+        inputs = processor(images=image, return_tensors="pt")
+        
+        # Make prediction
+        with torch.no_grad():
+            outputs = model(**inputs)
+            predictions = torch.nn.functional.softmax(outputs.logits, dim=-1)
+            predicted_class = predictions.argmax().item()
+        
+        print(f"Predicted class: {predicted_class}")
+        Dataset
+        This model was trained on the EuroSAT-SAR dataset, which contains Synthetic Aperture Radar (SAR) satellite images from the Sentinel-1 satellite. The dataset includes land use and land cover classification of European landscapes.
+        
+        Performance
+        The model achieves competitive performance on the EuroSAT-SAR test set. Detailed evaluation metrics can be found in the training logs.
+        
+        Training Infrastructure
+        Framework: PyTorch with Transformers
+        Hardware: CUDA-compatible GPU
+        Mixed Precision: Enabled for efficient training
+        Monitoring: TensorBoard for training visualization
+        Citation
+        If you use this model, please cite:
+        
+        BibTeX
+        @article{eurosat,
+          title={EuroSAT: A Novel Dataset and Deep Learning Benchmark for Land Use and Land Cover Classification},
+          author={Helber, Patrick and Bischke, Benjamin and Dengel, Andreas and Borth, Damian},
+          journal={IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing},
+          year={2019}
+        }
+        License
+        This model is released under the Apache 2.0 License. 

config.json

@@ -0,0 +1,35 @@
+{
+  "architectures": [
+    "EuroSATTransformerClassifier"
+  ],
+  "model_type": "eurosat-transformer",
+  "num_labels": 10,
+  "id2label": {
+    "0": "AnnualCrop",
+    "1": "Forest",
+    "2": "HerbaceousVegetation",
+    "3": "Highway",
+    "4": "Industrial",
+    "5": "Pasture",
+    "6": "PermanentCrop",
+    "7": "Residential",
+    "8": "River",
+    "9": "SeaLake"
+  },
+  "label2id": {
+    "AnnualCrop": 0,
+    "Forest": 1,
+    "HerbaceousVegetation": 2,
+    "Highway": 3,
+    "Industrial": 4,
+    "Pasture": 5,
+    "PermanentCrop": 6,
+    "Residential": 7,
+    "River": 8,
+    "SeaLake": 9
+  },
+  "base_model": "microsoft/swin-base-patch4-window7-224",
+  "image_size": 224,
+  "dropout_rate": 0.1,
+  "problem_type": "single_label_classification"
+}

evaluation_results.json

@@ -0,0 +1,150 @@
+{
+  "metrics": {
+    "accuracy": 0.9881481481481481,
+    "precision_weighted": 0.9882860757948263,
+    "recall_weighted": 0.9881481481481481,
+    "f1_weighted": 0.9881731188401808,
+    "precision_per_class": [
+      0.9800664451827242,
+      1.0,
+      0.9830508474576272,
+      0.9920318725099602,
+      1.0,
+      0.9898477157360406,
+      0.9498069498069498,
+      0.9966666666666667,
+      0.9920318725099602,
+      0.9966666666666667
+    ],
+    "recall_per_class": [
+      0.9833333333333333,
+      0.99,
+      0.9666666666666667,
+      0.996,
+      0.996,
+      0.975,
+      0.984,
+      0.9966666666666667,
+      0.996,
+      0.9966666666666667
+    ],
+    "f1_per_class": [
+      0.9816971713810315,
+      0.9949748743718593,
+      0.9747899159663865,
+      0.9940119760479043,
+      0.9979959919839679,
+      0.982367758186398,
+      0.9666011787819253,
+      0.9966666666666667,
+      0.9940119760479043,
+      0.9966666666666667
+    ],
+    "support_per_class": [
+      300,
+      300,
+      300,
+      250,
+      250,
+      200,
+      250,
+      300,
+      250,
+      300
+    ],
+    "classification_report": {
+      "AnnualCrop": {
+        "precision": 0.9800664451827242,
+        "recall": 0.9833333333333333,
+        "f1-score": 0.9816971713810315,
+        "support": 300
+      },
+      "Forest": {
+        "precision": 1.0,
+        "recall": 0.99,
+        "f1-score": 0.9949748743718593,
+        "support": 300
+      },
+      "HerbaceousVegetation": {
+        "precision": 0.9830508474576272,
+        "recall": 0.9666666666666667,
+        "f1-score": 0.9747899159663865,
+        "support": 300
+      },
+      "Highway": {
+        "precision": 0.9920318725099602,
+        "recall": 0.996,
+        "f1-score": 0.9940119760479043,
+        "support": 250
+      },
+      "Industrial": {
+        "precision": 1.0,
+        "recall": 0.996,
+        "f1-score": 0.9979959919839679,
+        "support": 250
+      },
+      "Pasture": {
+        "precision": 0.9898477157360406,
+        "recall": 0.975,
+        "f1-score": 0.982367758186398,
+        "support": 200
+      },
+      "PermanentCrop": {
+        "precision": 0.9498069498069498,
+        "recall": 0.984,
+        "f1-score": 0.9666011787819253,
+        "support": 250
+      },
+      "Residential": {
+        "precision": 0.9966666666666667,
+        "recall": 0.9966666666666667,
+        "f1-score": 0.9966666666666667,
+        "support": 300
+      },
+      "River": {
+        "precision": 0.9920318725099602,
+        "recall": 0.996,
+        "f1-score": 0.9940119760479043,
+        "support": 250
+      },
+      "SeaLake": {
+        "precision": 0.9966666666666667,
+        "recall": 0.9966666666666667,
+        "f1-score": 0.9966666666666667,
+        "support": 300
+      },
+      "accuracy": 0.9881481481481481,
+      "macro avg": {
+        "precision": 0.9880169036536595,
+        "recall": 0.9880333333333333,
+        "f1-score": 0.987978417610071,
+        "support": 2700
+      },
+      "weighted avg": {
+        "precision": 0.9882860757948263,
+        "recall": 0.9881481481481481,
+        "f1-score": 0.9881731188401808,
+        "support": 2700
+      }
+    }
+  },
+  "config": {
+    "model_name": "microsoft/swin-base-patch4-window7-224",
+    "num_labels": 10,
+    "batch_size": 32,
+    "learning_rate": 5e-05,
+    "num_epochs": 10
+  },
+  "label_names": [
+    "AnnualCrop",
+    "Forest",
+    "HerbaceousVegetation",
+    "Highway",
+    "Industrial",
+    "Pasture",
+    "PermanentCrop",
+    "Residential",
+    "River",
+    "SeaLake"
+  ]
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33db89561cc5e5a58faf3254d4d0194627a9e280c2bcb37030f11b7936569559
+size 350281670

training_args.json

@@ -0,0 +1,8 @@
+{
+  "learning_rate": 5e-05,
+  "num_epochs": 10,
+  "batch_size": 32,
+  "warmup_steps": 500,
+  "weight_decay": 0.01,
+  "model_name": "microsoft/swin-base-patch4-window7-224"
+}