The following repo contains the code for implementing the LeNet 5 Architechture.
The LeNet Architechture was discovered way back in the 1980's to recognise the postal codes on the MNIST Data set. The Details about the model can be found in the following paper :http://vision.stanford.edu/cs598_spring07/papers/Lecun98.pdf
In the Following repository i had implemented the LeNet 5 Architechture in two Datasets. I had used two data sets for the same namely:
- The MNIST Data set.
- The Fashion MNIST Data Set.
The MNIST Data Set contains Hand Written Digits from 0-9 which are process by the LeNet 5 CNN Architechture and then outputs the digit.
Obtained :
Training Accuracy - 99%
Test Accuracy - 98.02%
The Fashion MNIST Data set had 10 kinds of fashion objects which were fed as an input to the LeNet 5
Obtained :
Training Accuracy - 90%
Test Accuracy - 88%.
The Data sets for the models have not been provided in the repository as these data sets are inherently present in the Keras API.
MNIST DATA SET: https://images.app.goo.gl/BKZtTNSkuMCGZsyx8
Fashion MNIST DATA SET: https://images.app.goo.gl/q5r4z2aKF9fNLugA8