README.md

---
title: Animal Image Classification Using InceptionV3
emoji: 🐍
colorFrom: blue
colorTo: red
sdk: gradio
sdk_version: 6.5.1
app_file: app.py
pinned: false
license: mit
short_description: InceptionV3-based animal image classifier with data cleaning
models:
  - AIOmarRehan/Animal-Image-Classification-Using-CNN
datasets:
  - AIOmarRehan/AnimalsDataset
---

[If you would like a detailed explanation of this project, please refer to the Medium article below.](https://medium.com/@ai.omar.rehan/building-a-clean-reliable-and-accurate-animal-classifier-using-inceptionv3-175f30fbe6f3)

---
# Animal Image Classification Using InceptionV3

*A complete end-to-end pipeline for building a clean, reliable deep-learning classifier.*

This project implements a **full deep-learning workflow** for classifying animal images using **TensorFlow + InceptionV3**, with a major focus on **dataset validation and cleaning**. Before training the model, I built a comprehensive system to detect corrupted images, duplicates, brightness/contrast issues, mislabeled samples, and resolution outliers.

This repository contains the full pipeline—from dataset extraction to evaluation and model saving.

---

## Features

### Full Dataset Validation

The project includes automated checks for:

* Corrupted or unreadable images
* Hash-based duplicate detection
* Duplicate filenames
* Misplaced or incorrectly labeled images
* File naming inconsistencies
* Extremely dark/bright images
* Very low-contrast (blank) images
* Outlier resolutions

### Preprocessing & Augmentation

* Resize to 256×256
* Normalization
* Light augmentation (probabilistic)
* Efficient `tf.data` pipeline with caching, shuffling, prefetching

### Transfer Learning with InceptionV3

* Pretrained ImageNet weights
* Frozen base model
* Custom classification head (GAP → Dense → Dropout → Softmax)
* EarlyStopping + ModelCheckpoint + ReduceLROnPlateau callbacks

### Clean & Reproducible Training

* 80% training
* 10% validation
* 10% test
* High stability due to dataset cleaning

---

## 1. Dataset Extraction

The dataset is stored as a ZIP file (Google Drive). After mounting the drive, it is extracted and indexed into a Pandas DataFrame:

```python
drive.mount('/content/drive')

zip_path = '/content/drive/MyDrive/Animals.zip'
extract_to = '/content/my_data'

with zipfile.ZipFile(zip_path, 'r') as zip_ref:
    zip_ref.extractall(extract_to)
```

Each image entry records:

* Class
* Filename
* Full path

---

## 2. Dataset Exploration

Before any training, I analyzed:

* Class distribution
* Image dimensions
* Grayscale vs RGB
* Unique sizes
* Folder structures

Example class-count visualization:

```python
plt.figure(figsize=(32, 16))
class_count.plot(kind='bar')
```

This revealed imbalance and inconsistent image sizes early.

---

## 3. Visual Sanity Checks

Random images were displayed with their brightness, contrast, and shape to manually confirm dataset quality.

This step prevents hidden issues—especially in community-created or scraped datasets.

---

## 4. Data Quality Detection

The system checks for:

### Duplicate Images (Using MD5 Hashing)

```python
def get_hash(path):
    with open(path, 'rb') as f:
        return hashlib.md5(f.read()).hexdigest()

df['file_hash'] = df['full_path'].apply(get_hash)
duplicate_hashes = df[df.duplicated('file_hash', keep=False)]
```

### Corrupted Files

```python
try:
    with Image.open(file_path) as img:
        img.verify()
except:
    corrupted_files.append(file_path)
```

### Brightness/Contrast Outliers

Using PIL’s `ImageStat` to detect very dark/bright samples.

### Label Consistency Check

```python
folder = os.path.basename(os.path.dirname(row["full_path"]))
```

This catches mislabeled entries where folder name ≠ actual class.

---

## 5. Preprocessing Pipeline

Custom preprocessing:

* Resize → Normalize
* Optional augmentation
* Efficient `tf.data` batching

```python
def preprocess_image(path, target_size=(256, 256), augment=True):
    img = tf.image.decode_image(...)
    img = tf.image.resize(img, target_size)
    img = img / 255.0
```

Split structure:

| Split      | Percent |
| ---------- | ------- |
| Train      | 80%     |
| Validation | 10%     |
| Test       | 10%     |

---

## 6. Model — Transfer Learning with InceptionV3

The model is built using **InceptionV3 pretrained on ImageNet** as a feature extractor.

```python
inception = InceptionV3(
    input_shape=input_shape,
    weights="imagenet",
    include_top=False
)
```

At first, **all backbone layers are frozen** to preserve pretrained representations:

```python
for layer in inception.layers:
    layer.trainable = False
```

A custom classification head is added:

* GlobalAveragePooling2D
* Dense(512, ReLU)
* Dropout(0.5)
* Dense(N, Softmax) — where *N = number of classes*

This setup allows the model to learn dataset-specific patterns while avoiding overfitting during early training.

---

## 7. Initial Training (Frozen Backbone "Feature Extraction")

The model is compiled using:

* **Loss**: `sparse_categorical_crossentropy`
* **Optimizer**: Adam
* **Metric**: Accuracy

Training is performed with callbacks to improve stability:

* **EarlyStopping** (restore best weights)
* **ModelCheckpoint** (save best model)
* **ReduceLROnPlateau**

```python
history = model.fit(
    train_ds,
    validation_data=val_ds,
    epochs=5,
    callbacks=callbacks
)
```

This stage allows the new classification head to converge while keeping the pretrained backbone intact.

---

## 8. Fine-Tuning the InceptionV3 Backbone

After the initial convergence, **fine-tuning is applied** to improve performance.

The **last 30 layers** of InceptionV3 are unfrozen:

The model is then recompiled and trained again:

```python
history = model.fit(
    train_ds,
    validation_data=val_ds,
    epochs=10,
    callbacks=callbacks
)
```

Fine-tuning allows higher-level convolutional filters to adapt to the animal dataset, resulting in better class separation and generalization.

---

## 9. Model Evaluation

The final model is evaluated on a **held-out test set**.

### Accuracy & Loss Curves

Training and validation curves are plotted to monitor:

* Convergence behavior
* Overfitting
* Generalization stability

These plots confirm that fine-tuning improves validation performance without introducing instability.

![Charts](https://files.catbox.moe/qatkd6.png)

---

### Confusion Matrix

A confusion matrix is computed to visualize class-level performance:

* Highlights misclassification patterns
* Reveals class confusion (e.g., visually similar animals)

Both annotated and heatmap-style confusion matrices are generated.

![Confusion Matrix](https://files.catbox.moe/ih64qj.png)

---

### Classification Metrics

The following metrics are computed on the test set:

* **Accuracy**
* **Precision (macro)**
* **Recall (macro)**
* **F1-score (macro)**

A detailed **per-class classification report** is also produced:

* Precision
* Recall
* F1-score
* Support

This provides a deeper understanding beyond accuracy alone.

```
10/10 ━━━━━━━━━━━━━━━━━━━━ 1s 106ms/step - accuracy: 0.9826 - loss: 0.3082 - Test Accuracy: 0.9900
10/10 ━━━━━━━━━━━━━━━━━━━━ 1s 93ms/step

Classification Report:
              precision    recall  f1-score   support

        cats       0.99      0.97      0.98       100
        dogs       0.97      0.99      0.98       100
      snakes       1.00      1.00      1.00       100

    accuracy                           0.99       300
   macro avg       0.99      0.99      0.99       300
weighted avg       0.99      0.99      0.99       300
```

---

### ROC Curves (Multi-Class)

To further evaluate model discrimination:

* One-vs-Rest ROC curves are generated per class
* A **macro-average ROC curve** is computed
* AUC is reported for overall performance

These curves demonstrate strong separability across all classes.

![ROC Curve](https://files.catbox.moe/a8aaa5.png)

---

## 10. Final Model

The best-performing model (after fine-tuning) is saved and later used for deployment:

```python
model.save("Inception_V3_Animals_Classification.h5")
```

This trained model is deployed using **FastAPI + Docker** for inference and **Gradio on Hugging Face Spaces** for public interaction.

---

## Updated Key Takeaways

> **Clean data enables strong fine-tuning.**

By combining:

* Rigorous dataset validation
* Transfer learning
* Selective fine-tuning
* Comprehensive evaluation

the model achieves **high accuracy, stable convergence, and reliable real-world performance**.

Fine-tuning only a subset of pretrained layers strikes the optimal balance between **generalization and specialization**.

---