{
  "id": 200812,
  "title": "[Pet project]: Traffic lights modelling",
  "url": "/competitions/lyft-motion-prediction-autonomous-vehicles/discussion/200812",
  "author_name": "",
  "post_date": "2020-12-02T01:34:45.964292700Z",
  "votes": 4,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hi Everyone,</p>\n<p><em>First, I'd like to thank the organizers for the interesting dataset, <a href=\"https://www.kaggle.com/iglovikov\" target=\"_blank\">@iglovikov</a> for publishing valuable tips helping me to grow (yet-another valuable tips!), and thank everyone in the community for publishing and describing the amazing solutions!</em></p>\n<hr>\n<h1>Pet Project Questions</h1>\n<p>I looked into predicting the status of all traffic lights at an intersection given limited SDV's observations. </p>\n<p>Basically, I got curious about the following questions:</p>\n<ul>\n<li><p>when observing just a single red traffic light, how well can we guess the current color of all other traffic lights at the intersection?<br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F493138%2Ff2aeb8a35471c92da24128084e19931e%2Fopen_question.jpg?generation=1606871916180248&amp;alt=media\" alt=\"\"></p></li>\n<li><p>How would our current best guesses change if in addition to the red traffic light we observe some vehicles at the intersection?</p></li>\n</ul>\n<hr>\n<h1>Answers</h1>\n<p>In my experiments, for this particular scene I got the following out-of-sample guesses:</p>\n<ul>\n<li><p>After observing a single red light<br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F493138%2F88e91543e2351e6a3591d53150968bc2%2Fanswer_1.jpg?generation=1606868157937497&amp;alt=media\" alt=\"\"></p></li>\n<li><p>After observing some vehicles in addition to the red traffic light:<br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F493138%2F29f72eea8952c20f5d83dd1d1c6bf453%2Fanswer_2.jpg?generation=1606868519360968&amp;alt=media\" alt=\"\"></p></li>\n<li><p>Later, after observing more vehicles for a longer period of time:<br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F493138%2Fd570e3c60584bbe58b913ae6df4c39af%2Fanswer_3.jpg?generation=1606868672364929&amp;alt=media\" alt=\"\"></p></li>\n</ul>\n<hr>\n<h1><a href=\"https://github.com/SamusRam/traffic_lights\" target=\"_blank\">GitHub repo</a></h1>\n<p>In case anyone might be interested, more details can be found in <a href=\"https://github.com/SamusRam/traffic_lights\" target=\"_blank\">the GitHub repo of this pet project</a>.</p>",
  "messages": [
    {
      "id": "1098964",
      "postDate": "12/02/2020 01:34:45",
      "content": "<p>Hi Everyone,</p>\n<p><em>First, I'd like to thank the organizers for the interesting dataset, <a href=\"https://www.kaggle.com/iglovikov\" target=\"_blank\">@iglovikov</a> for publishing valuable tips helping me to grow (yet-another valuable tips!), and thank everyone in the community for publishing and describing the amazing solutions!</em></p>\n<hr>\n<h1>Pet Project Questions</h1>\n<p>I looked into predicting the status of all traffic lights at an intersection given limited SDV's observations. </p>\n<p>Basically, I got curious about the following questions:</p>\n<ul>\n<li><p>when observing just a single red traffic light, how well can we guess the current color of all other traffic lights at the intersection?<br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F493138%2Ff2aeb8a35471c92da24128084e19931e%2Fopen_question.jpg?generation=1606871916180248&amp;alt=media\" alt=\"\"></p></li>\n<li><p>How would our current best guesses change if in addition to the red traffic light we observe some vehicles at the intersection?</p></li>\n</ul>\n<hr>\n<h1>Answers</h1>\n<p>In my experiments, for this particular scene I got the following out-of-sample guesses:</p>\n<ul>\n<li><p>After observing a single red light<br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F493138%2F88e91543e2351e6a3591d53150968bc2%2Fanswer_1.jpg?generation=1606868157937497&amp;alt=media\" alt=\"\"></p></li>\n<li><p>After observing some vehicles in addition to the red traffic light:<br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F493138%2F29f72eea8952c20f5d83dd1d1c6bf453%2Fanswer_2.jpg?generation=1606868519360968&amp;alt=media\" alt=\"\"></p></li>\n<li><p>Later, after observing more vehicles for a longer period of time:<br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F493138%2Fd570e3c60584bbe58b913ae6df4c39af%2Fanswer_3.jpg?generation=1606868672364929&amp;alt=media\" alt=\"\"></p></li>\n</ul>\n<hr>\n<h1><a href=\"https://github.com/SamusRam/traffic_lights\" target=\"_blank\">GitHub repo</a></h1>\n<p>In case anyone might be interested, more details can be found in <a href=\"https://github.com/SamusRam/traffic_lights\" target=\"_blank\">the GitHub repo of this pet project</a>.</p>",
      "rawMarkdown": "Hi Everyone,\n\n*First, I'd like to thank the organizers for the interesting dataset, @iglovikov for publishing valuable tips helping me to grow (yet-another valuable tips!), and thank everyone in the community for publishing and describing the amazing solutions!*\n_____________________________\n# Pet Project Questions\n\nI looked into predicting the status of all traffic lights at an intersection given limited SDV's observations. \n\nBasically, I got curious about the following questions:\n- when observing just a single red traffic light, how well can we guess the current color of all other traffic lights at the intersection?\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F493138%2Ff2aeb8a35471c92da24128084e19931e%2Fopen_question.jpg?generation=1606871916180248&alt=media)\n\n- How would our current best guesses change if in addition to the red traffic light we observe some vehicles at the intersection?\n____________________________\n# Answers\nIn my experiments, for this particular scene I got the following out-of-sample guesses:\n  * After observing a single red light\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F493138%2F88e91543e2351e6a3591d53150968bc2%2Fanswer_1.jpg?generation=1606868157937497&alt=media)\n\n  * After observing some vehicles in addition to the red traffic light:\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F493138%2F29f72eea8952c20f5d83dd1d1c6bf453%2Fanswer_2.jpg?generation=1606868519360968&alt=media)\n\n  * Later, after observing more vehicles for a longer period of time:\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F493138%2Fd570e3c60584bbe58b913ae6df4c39af%2Fanswer_3.jpg?generation=1606868672364929&alt=media)\n_____________________________\n#  [GitHub repo](https://github.com/SamusRam/traffic_lights)\nIn case anyone might be interested, more details can be found in [the GitHub repo of this pet project](https://github.com/SamusRam/traffic_lights).",
      "votes": null
    },
    {
      "id": "1512823",
      "postDate": "09/14/2021 16:06:09",
      "content": "<p>Can it be possible to get the best green times in that same intersection?</p>",
      "rawMarkdown": "Can it be possible to get the best green times in that same intersection?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1512823,
      "author_name": "joshohomina",
      "author_url": "",
      "post_date": "09/14/2021 16:06:09",
      "content": "<p>Can it be possible to get the best green times in that same intersection?</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1098964": "Hi Everyone,\n\n*First, I'd like to thank the organizers for the interesting dataset, @iglovikov for publishing valuable tips helping me to grow (yet-another valuable tips!), and thank everyone in the community for publishing and describing the amazing solutions!*\n_____________________________\n# Pet Project Questions\n\nI looked into predicting the status of all traffic lights at an intersection given limited SDV's observations. \n\nBasically, I got curious about the following questions:\n- when observing just a single red traffic light, how well can we guess the current color of all other traffic lights at the intersection?\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F493138%2Ff2aeb8a35471c92da24128084e19931e%2Fopen_question.jpg?generation=1606871916180248&alt=media)\n\n- How would our current best guesses change if in addition to the red traffic light we observe some vehicles at the intersection?\n____________________________\n# Answers\nIn my experiments, for this particular scene I got the following out-of-sample guesses:\n  * After observing a single red light\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F493138%2F88e91543e2351e6a3591d53150968bc2%2Fanswer_1.jpg?generation=1606868157937497&alt=media)\n\n  * After observing some vehicles in addition to the red traffic light:\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F493138%2F29f72eea8952c20f5d83dd1d1c6bf453%2Fanswer_2.jpg?generation=1606868519360968&alt=media)\n\n  * Later, after observing more vehicles for a longer period of time:\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F493138%2Fd570e3c60584bbe58b913ae6df4c39af%2Fanswer_3.jpg?generation=1606868672364929&alt=media)\n_____________________________\n#  [GitHub repo](https://github.com/SamusRam/traffic_lights)\nIn case anyone might be interested, more details can be found in [the GitHub repo of this pet project](https://github.com/SamusRam/traffic_lights).",
    "1512823": "Can it be possible to get the best green times in that same intersection?"
  },
  "source": "meta"
}