Update README.md

README.md

@@ -1,3 +1,28 @@
+---
+language: en
+tags:
+- image-classification
+- CNN
+- Convolution Neural Entwork
+- Nueral Network
+- Trash
+metrics:
+- name: train-accuracy
+  value: 91%
+- name: test-accuracy
+  value: 55%
+pipeline:
+  - image-classification
+libraries:
+  - name: torch
+    version: 1.9.0
+  - name: torchvision
+    version: 0.10.0
+  - name: numpy
+    version: 1.21.0
+---
+
+
 ## Trash Classification CNN Model
 
 ### What is this?
@@ -55,10 +80,6 @@ Additionally, the repository contains 2 folders:
 - **samples**: This has 10 pictures, you can use for testing the model using `predict.py`.
 
 
-### What I Learned?
-
-This project taught me the basics of **Computer Vision** with **PyTorch**, a lot about **Convolutional Neural Networks (CNNs)**, and also taught me how to **model** my project. It also taught me how to write **readable code** and handle **errors**, especially in the `predict.py` file.
-
-I gained understanding about **classification** and how to implement it with **neural networks** and **deep learning**. While working on this, I learned the basics of **matplotlib** and **mlxtend** and also realized the impact of **data quantity** on results, which led to the decision of using only **30% of the data**.
+## Model Overview
 
-I found that the best working **optimizer** with **TinyVGG** was an **Adam Optimizer** with a **learning rate** of **0.001**, trained on **20 epochs** and a **batch size** of **32** with **15** **hidden units**. This resulting in **Train Loss** of _**0.24**_ and **Test Loss** of _**2.17**_, **Train Accuracy** of _**91%**_ and **Test Accuracy** of _**55%**_.   
+This model is designed for image classification tasks. It requires input images of size 112x112 pixels.