{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Deep Learning for Computer Vision:  Final Project"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Computer Science: COMS W 4995 007"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Proposal: November 12, 2019\n",
    "### Presentations: December 3 and 5, 2019\n",
    "### Report: December 10, 2019"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Project Overview"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The final project is one of the most important and, hopefully, exciting components of the course. You will have the opportunity to develop a deep learning system of your own choosing. \n",
    "You are free to select whatever framework (Pytorch, Tensorflow) or wrapper (Keras, TFLearn) you like, but you need create a report on your project in a Jupyter notebook. You are also free build on publically available models and code, but your report must clearly give attribution for the work of others and must clearly delineate your contributions. Also, half of the class will present their project during the last 2 days. All of the class will prepare videos of their presentation and submit these when the final report is due. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "### Project Proposal"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The project description should include the title of the project, participants, a description of the objectives of the project, and a plan for how the project will be completed. The description of the objectives should include modest predictions of the success of the project. The plan for completion should include a description of the training data and how it will be obtained, a discussion of what deep learning framework will be used and why, and a rough description of the planned network architecture.\n",
    "\n",
    "You are permitted to work together on a project in groups of two or three, but group size must not exceed three participants.  For group projects there must be a clearly delineated division of labor: you should state in the project description and project report who was responsible for which portion of the project. Each student must hand in a separate report. (Students will not necessarily get the same grade for the same project.)\n",
    "\n",
    "You should mention whether you are simply re-implementing what others have done before but applying to new data or whether you are attempting to do something new to the best of your knowledge. Creative and original projects will be judged more kindly than those that are rehashing something in the existing literature. And projects that include a component in which data is acquired/curated into training and validation sets will be veiwed more favorably than those that simply download an existing data set such as CIFAR-100.\n",
    "\n",
    "As this is a computer vision course it is expected that your data will be visual, but excpetions might be made if the student is enthusiatic and persuasive enough. The most straightforward project would be to build a system that classifies images into catogories. A more difficult project might be to build a system that detects and localizes a type of object within an image. A still more complicated project might involve joining a ConvNet with an LSTM for a problem (like image captioning) that requires vision and language. But again, creative and original projects will be judged more kindly.  \n",
    "\n",
    "It is important to scope your project so that you get some working results. Project reports that say \"I tried this and this but nothing seemed to work...\" are discouraged. \n",
    "Above all, you should demonstrate end-to-end fluency in the basics of deep learning. \n",
    "\n",
    "I cannot wait to see the results. Good luck!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Project Presentations"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "To allow students to present their work in two class periods, each student will have only 3 minutes, not a second more. We will be strict about the timing, so you should practice your presentation. The key here is to get across three things: what you did, how you did it, and how well it worked. Students working in groups of two will get 6 minutes and groups of three will get 9 minutes. Note only half of the class groups will present during the last two days of class, but all of the groups will submit videos of their project. The time limit rules for the videos is the same as for the presentations. And the video can simply be a narration over a slideshow, but again each student in a group needs to present/narrate the work they did.  \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Project Reports"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The report should be done as a Jupyter Notebook. The report should be a complete description of the objectives of the work, the methods used to solve the problem, experimental evidence of a working system, the code, and clear delineation of what you have done vs. what you are leveraging that others have done. If you have used the work of others YOU MUST INCLUDE ATTRIBUTION by citing this work inline and as part of a \"bibliography\" at the end. You should describe what worked, what did not, and why. If you are working in a group you need to submit your own report and this report should be clear about what your individual contribution was. It is ok to include your collaborators work in your report, but you must be clear about your section and write this yourself."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}