Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,52 @@
|
|
1 |
-
---
|
2 |
-
license: apache-2.0
|
3 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: apache-2.0
|
3 |
+
---
|
4 |
+
**Real art vs AI-Generated art image classification**
|
5 |
+
|
6 |
+
This project provides a ResNet50 pre-trained model for classifying images as either 'real art' or 'fake art'.
|
7 |
+
ResNet50 is a deep convolutional neural network with 50 layers, known for its "residual connections" that help mitigate the vanishing gradient problem.
|
8 |
+
It allows training of very deep networks by adding shortcut connections that skip one or more layers, making it highly effective for image classification tasks.
|
9 |
+
Our goal is to accurately classify the source of the image with at least 85% accuracy and achieve at least 80% in the Recall test.
|
10 |
+
|
11 |
+
*Installation instructions*
|
12 |
+
|
13 |
+
The following libraries or packages are required: numpy, pandas, tensorflow, keras, matplotlib, sklearn, cv2.
|
14 |
+
We prepare the data for the model by sorted the images into 2 types of folders which are divided equally(real art- labeled as 0, fake art- labeled as 1).
|
15 |
+
Our ResNet50 model is based on 2,800 images that have been resized and normalized, the files formats is PNG, JPG.
|
16 |
+
The images are divided into a training set that contains 90% from data and a testing set that contains the remaining 10%.
|
17 |
+
|
18 |
+
*ResNet50 model architecture*
|
19 |
+
|
20 |
+
The model is pre-trained on 'ImageNet' that contains a large dataset of more than millions images.
|
21 |
+
It applies transfer learning, freezing initial layers of ResNet50, and training only the final layers.
|
22 |
+
The final layer, which makes the predictions, is a binary classification layer that uses a sigmoid activation function.
|
23 |
+
|
24 |
+
*Training Details*
|
25 |
+
|
26 |
+
The model is trained using binary cross-entropy loss and the Adam optimizer.
|
27 |
+
The model validates itself during training using 20% of the training data as validation, independent of the test data, to monitor performance and avoid overfitting.
|
28 |
+
The model is trained for 5 epochs with a batch size of 32 and employs 4-fold cross-validation to ensure robust performance.
|
29 |
+
During each fold, the model's weights are saved after training, allowing for the reuse of the best-performing weights.
|
30 |
+
|
31 |
+
*Performance Evaluation*
|
32 |
+
|
33 |
+
After training, the model is evaluated on the test set.
|
34 |
+
The following metrics are used to measure performance:
|
35 |
+
Accuracy: The percentage of correct classifications.
|
36 |
+
Precision, Recall, F1-Score: For evaluating the model’s classification ability on both real art and AI-generated art images.
|
37 |
+
Confusion Matrix: Provides insights into classification performance. Displays true positives, false positives, true negatives, and false negatives.
|
38 |
+
|
39 |
+
*To run the project*
|
40 |
+
|
41 |
+
Place the images in the respective training and testing folders.
|
42 |
+
Preprocess the images by resizing and normalizing them.
|
43 |
+
Train the model using the provided code.
|
44 |
+
Evaluate the model on the test set.
|
45 |
+
|
46 |
+
*Visualization results*
|
47 |
+
|
48 |
+
Confusion Matrix: To visualize the classification performance.
|
49 |
+
Training and Validation Metrics: Plots for accuracy and loss over the epochs.
|
50 |
+
|
51 |
+
|
52 |
+
|