https://github.com/jlenon7/cifar_image_model

🤖 Model that predicts from which category an image belongs (cat, bird, airplane, etc...).
https://github.com/jlenon7/cifar_image_model

Last synced: 25 days ago
JSON representation

🤖 Model that predicts from which category an image belongs (cat, bird, airplane, etc...).

• Host: GitHub
• URL: https://github.com/jlenon7/cifar_image_model
• Owner: jlenon7
• License: mit
• Created: 2024-06-06T14:24:03.000Z (7 months ago)
• Default Branch: main
• Last Pushed: 2024-06-06T14:36:38.000Z (7 months ago)
• Last Synced: 2024-11-27T06:47:41.906Z (26 days ago)
• Language: Python
• Size: 48.6 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Cifar Image Model 🤖

> Convolutional neural network that predicts from which category an image belongs.

> The model was trained using Cifar10 dataset.

## Goals

Predict from which category an image belongs:



## Results

### Confusion matrix

```shell

[[734  19  42  25  29  13   8  22  65  43]

 [ 27 803  11  15   5  11   3  13  26  86]

 [ 65   8 516 109 102 112  22  43  17   6]

 [ 22   9  57 507  83 223  30  42  14  13]

 [ 16   3  51  77 655  67  18 101  10   2]

 [ 13   1  51 181  51 629  10  49   8   7]

 [  8   4  70  92  81  59 657  13  10   6]

 [ 19   1  25  49  60  79   5 741   7  14]

 [ 62  30  18  18  14  13   2   6 814  23]

 [ 46  59  13  34   8  10   7  18  32 773]]

```

### Classification report

```shell

              precision    recall  f1-score   support

           0       0.73      0.73      0.73      1000

           1       0.86      0.80      0.83      1000

           2       0.60      0.52      0.56      1000

           3       0.46      0.51      0.48      1000

           4       0.60      0.66      0.63      1000

           5       0.52      0.63      0.57      1000

           6       0.86      0.66      0.75      1000

           7       0.71      0.74      0.72      1000

           8       0.81      0.81      0.81      1000

           9       0.79      0.77      0.78      1000

    accuracy                           0.68     10000

   macro avg       0.69      0.68      0.69     10000

weighted avg       0.69      0.68      0.69     10000

```

## Exploratory data analysis

### Loss



### Accuracy



### Heatmap Confusion Matrix



## Running

To run the model first create a new Python environment and activate it. I'm using [Anaconda](https://www.anaconda.com/) for setting the python version that pipenv should use to set up the environment. The command bellow will automatically setup the environment with conda and pipenv:

```shell

make env

```

Now install all the project dependencies:

```shell

make install-all

```

And run the model:

```shell

make model

```

> [!WARNING]

> Dont run `make model` without deleting `storage/cifar-image-model.keras`, this will

> cause train/test data over fitting.

After running you model, it will be saved inside `storage/cifar-image-model.keras`.

To just run your recent created model and predict a random value from our data set,

use the following script:

```shell

make predict 

```

> [!WARNING]

> In case you have deleted the `storage/cifar-image-model.keras`, remember that to get `make predict` working you need to run `make model` first to create it.

To run TensorBoard with the latest created version of the model within this

repository run:

```shell

make board

```