This data is provided by Microsoft on 5/9/2022.
ref-https://www.microsoft.com/en-us/download/details.aspx?id=54765
- After data Collection we need to segreate the data based on classification before feeding to the model.
- Imagaes we have are of a higher resolution hence while building a model we need to pass it through various filters and pooling to lower the resolution and keep the important features unharmed.
- Once images get to the required resolution we can Flatten them and send them to Dense layer for Learning.
- As there are only 2 categories we wil choose to use Sigmoid Function as a output layer.
Construct a CNN model that achieves a a accuracy of greater than 80% meanwhile trying to keep the epoochs low.