Multi-classification Problem Lesson 1

diazagasatya · October 30, 2018, 9:36pm

Hi, this is my implementation of lesson one using resnet34 to classify planet surfaces.

diazagasatya/deep_learning_fastai/blob/master/main/lesson1-planets-dagasatya.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Lesson 1 - CNN Image Classification\n",
    "## Diaz Agasatya - Data Institute Deep Learning Part 1\n",
    "The general outcome of this lesson is to be able to run a 34 CNN resnet from fastai library on Google Compute Platform GPU to train a model to classify images.\n",
    "\n",
    "First of all every notebook should start which these following three lines. The first line will ensure to reload any libraries that will be used in this notebook, the second line will autoreload, and the last line will reload images and charts in the notebook.\n",
    "\n",
    "In this lesson, we will train the Resnet34 model to for a multiclassification planet dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],

This file has been truncated. show original

Problem:
Error rate is too high

ramesh · October 31, 2018, 1:38am

This dataset is in Lesson 2 Notebook. Check it out - https://github.com/fastai/course-v3/blob/master/nbs/dl1/lesson2-planet.ipynb

diazagasatya · October 31, 2018, 4:49pm

Hi Ramesh! Thank you for the link!

ricknta · November 7, 2018, 1:53am

Hey Diaz were you able to download the Kaggle planet data ok as in the lesson 2 nb?