https://github.com/p-mcc/image_classification_with_pytorch

Playing Cards Image Classification utilizing PyTorch Framework
https://github.com/p-mcc/image_classification_with_pytorch

alexnet deep-learning image-classification playing-cards python pytorch

Last synced: 3 months ago
JSON representation

Playing Cards Image Classification utilizing PyTorch Framework

• Host: GitHub
• URL: https://github.com/p-mcc/image_classification_with_pytorch
• Owner: P-MCC
• Created: 2024-11-13T11:49:39.000Z (11 months ago)
• Default Branch: main
• Last Pushed: 2024-11-21T23:42:13.000Z (11 months ago)
• Last Synced: 2024-11-22T00:24:25.938Z (11 months ago)
• Topics: alexnet, deep-learning, image-classification, playing-cards, python, pytorch
• Language: Python
• Homepage:
• Size: 442 KB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
#### Image Classification With Pytorch

This deep learning project focuses on classifying images of playing cards, utilizing the PyTorch framework. The chosen model architecture for this project is AlexNet (built from scratch, not using transfer learning).

**Training Details** :

* **Epochs** : 50

* **Batch Size** : 8

* **Learning Rate** : 0.0005, optimized using the Adam optimizer with a weight decay of `1e-4`.

**Data Preprocessing** :

* Images were resized to **227x227 pixels** as the sole transformation; no data augmentation techniques (e.g., flipping or rotation) were used.

**Validation Methods** :

* Validation was performed with Hold-Out Validation Method at the end of each epoch using the **validation dataset** to calculate loss and accuracy.

* It employs a **simple train-validation split** where the training and validation sets are predefined and separate. The training loop iterates through the `train_loader`, and after each epoch, the model is evaluated on the `valid_loader` to measure validation loss and accuracy.

### Results:

**Training results:**

![Graphs](Assets/graphs.png)

**Testing results:**

Test Accuracy: ***0.8830***

Classification Report:

|              | precision | recall | f1-score | support |

| ------------ | --------- | ------ | -------- | ------- |

| accuracy     |           |        | 0.88     | 265     |

| macro avg    | 0.91      | 0.88   | 0.88     | 265     |

| weighted avg | 0.91      | 0.88   | 0.88     | 265     |

Confusion Matrix 

![cf](Assets/cf.png)

**Custom Image Test Result:**

![custom](Assets/custom_playing_card_image_classification.png)