# Dynamic Vision Sensor

**URL:** <https://discourse.numenta.org/t/dynamic-vision-sensor/2075>\
**Category:** Engineering\
**Tags:** encoders\
**Created:** [March 31, 2017, 6:56am UTC](https://discourse.numenta.org/t/dynamic-vision-sensor/2075 "2017-03-31T06:56:27Z")\
**Posts on this page:** 5\
**Page:** 1

<div class="post-metadata">

**Author:** ![juanhorta](https://yyz2.discourse-cdn.com/flex030/user_avatar/discourse.numenta.org/juanhorta/32/831_2.png) [@juanhorta](https://discourse.numenta.org/u/juanhorta)\
**Post date:** [March 31, 2017, 6:56am UTC](https://discourse.numenta.org/t/dynamic-vision-sensor/2075/1 "2017-03-31T06:56:27Z")

</div>

I thought it could be inspiring for the development of sensors (more even for sensorimotor integration):

[http://rpg.ifi.uzh.ch/research\_dvs.html](http://rpg.ifi.uzh.ch/research_dvs.html)

---

<div class="post-metadata">

**Author:** ![jakebruce](https://yyz2.discourse-cdn.com/flex030/user_avatar/discourse.numenta.org/jakebruce/32/996_2.png) [@jakebruce](https://discourse.numenta.org/u/jakebruce)\
**Post date:** [April 2, 2017, 11:19pm UTC](https://discourse.numenta.org/t/dynamic-vision-sensor/2075/2 "2017-04-02T23:19:04Z")

</div>

We have a DVS camera in our lab, it’s very intriguing, especially the incredibly high effective framerates. One potential caveat is that the output is unbelievably noisy:

---

<div class="post-metadata">

**Author:** ![rhyolight](https://yyz2.discourse-cdn.com/flex030/user_avatar/discourse.numenta.org/rhyolight/32/3922_2.png) [@rhyolight](https://discourse.numenta.org/u/rhyolight)\
**Post date:** [April 3, 2017, 3:10pm UTC](https://discourse.numenta.org/t/dynamic-vision-sensor/2075/3 "2017-04-03T15:10:55Z")

</div>

Very noisy, but doesn’t look too noisy to to SDR comparison operations successfully!

---

<div class="post-metadata">

**Author:** ![sunguralikaan](https://yyz2.discourse-cdn.com/flex030/user_avatar/discourse.numenta.org/sunguralikaan/32/585_2.png) [@sunguralikaan](https://discourse.numenta.org/u/sunguralikaan)\
**Post date:** [September 5, 2017, 8:03pm UTC](https://discourse.numenta.org/t/dynamic-vision-sensor/2075/4 "2017-09-05T20:03:19Z")

</div>

@juanhorta, I recently started tinkering with this which seems more valuable the more I think on it, thanks for the idea. HTM models change in input. These sensors abstract the **change** in visual data in a sparse way.

---

<div class="post-metadata">

**Author:** ![fcred](https://yyz2.discourse-cdn.com/flex030/user_avatar/discourse.numenta.org/fcred/32/433_2.png) [@fcred](https://discourse.numenta.org/u/fcred)\
**Post date:** [September 19, 2017, 7:05am UTC](https://discourse.numenta.org/t/dynamic-vision-sensor/2075/5 "2017-09-19T07:05:30Z")

</div>

For those without the budget for a real DVS, this [library](https://github.com/chanokin/pyDVS) emulates the output of a DVS using a normal frame based camera or single images and movies as input.