5-2 15 Naga Venkata Satya Pranoop Mutha

5-2 23 Geovanni West

---

---

Using the inception model:

1. Train the given dataset(given in source code) for 100 iterations and report accuracy
2. Provide TensorBoardvisualizations for Training, weights, loss etc. and validation.

Inception-v3 is trained for the ImageNet Large Visual Recognition Challenge using the data from 2012. This is a standard task in computer vision, where models try to classify entire images into 1000 classes.

---

Adjusting the iterations to 100:

Final test accuracy was 87.3%

After changing to 200 iterations, the final test accuracy was 89.9%

Self-test: The first image saw a 91.5% match of tulips.

The second image saw a 49.6% match for roses and a 43.9% match for sunflowers.

Tensorflow visualisations are as follows. The data is based off the 200 iteration run:

Accuracy and Cross Entropy


Activation


Bias Mean


Weights Mean


Bias and weight Histogram