Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# AlexNet
|
| 2 |
+
|
| 3 |
+
Submission for coding assignment of Fatima Fellowship.
|
| 4 |
+
|
| 5 |
+
## Model Details
|
| 6 |
+
|
| 7 |
+
This is a simple implementation of AlexNet in Keras for the coding assignment of Fatima Fellowship.
|
| 8 |
+
|
| 9 |
+
### Model Description
|
| 10 |
+
|
| 11 |
+
- **Developed by:** Madhav Kumar
|
| 12 |
+
- **Model type:** Convolutional Neural Network
|
| 13 |
+
|
| 14 |
+
## How to Get Started with the Model
|
| 15 |
+
|
| 16 |
+
The files contains the notebook Alexnet_without_cropping.ipynb. It contains the code that was used to train the model. In the files, "weights-improvement-07-0.87.hdf5" contain the weights of the best model during training.
|
| 17 |
+
"logs-20230326T161429Z-001.zip" contain the logs of the training.
|
| 18 |
+
|
| 19 |
+
## Training Details
|
| 20 |
+
|
| 21 |
+
### Training Data
|
| 22 |
+
|
| 23 |
+
The files contain the training data images in train.zip and the data with image names and labels in the train.csv.
|
| 24 |
+
|
| 25 |
+
### Training Procedure
|
| 26 |
+
|
| 27 |
+
I trained the AlexNet model on the training data on Google Colab.
|
| 28 |
+
|
| 29 |
+
#### Preprocessing
|
| 30 |
+
|
| 31 |
+
I used data augmentaion to increase the size of training data.
|
| 32 |
+
|
| 33 |
+
## Evaluation
|
| 34 |
+
|
| 35 |
+
The model was tested on a validation data which was 20% percent of the train.zip data. The maximum accuracy obtained was 86.88%.
|