Thursday, January 21, 2021
Smartsens Claims #1 Spot in CIS Volume for Machine Vision Applications
Smartsens publishes a promotional video on global shutter advantages where it claims to be #1 in terms of machine vision image sensors shipment volume:
Update: Smartsens has updated the video with explanations on of the machine vision market positioning:
This is a broad claim. what do they count as 'machine vision'? I personally dont count drones to this market, isnt this more a 'consumer' type of product? I work in robotics/machine vision field and for sure dont have the full picture of the market... but I dont know of a single industrial camera for robotics (like the cmount style global shutter cameras that nowerdays often use Sony Pregius type of sensors) that uses a Smartsens sensor. Can you name a few such examples?
ReplyDeleteI mean... take this (recently updated) 'periodic table of machine vision image sensors' by Flir/PointGrey. https://www.flir.com/globalassets/discover/machine-vision/sensor-periodic-table/sensor-periodic-table_2021.pdf. This (the market the type of company like PointGrey serves) is more or less my market view on sensors for machine vision.
DeleteUpdated table is missing quite some new sensors, but including some 10+ year old devices.
Deletemaybe the 200m cameras used for mass surveillance in china?
ReplyDeleteIt's an easy claim, while they don't share figures...
ReplyDeleteEven including consumer global shutter we can doubt they are the first, looking at others
Surveillance cameras, feature extraction, face recognition ... scary stuff
ReplyDeleteYep, I guess we do not realize that China is a massive market in surveillance, ITS, etc, served by Chinese companies like SmartSens.
ReplyDeleteCongrats to SmartSens! Innovative and growing company
yes - but they explicitly say 'machine vision'. The global shutter sensors on their website have a property that does not match the 'machine vision' field I know - a quite noticable CRA. In 'machine vision' as I know it, this is a problem because often telecentric lens are used and you get vignetting with such sensors. The sensors I know most times have CRA (near) 0 for this reason.
DeleteSurveillance cameras, feature extraction, face recognition are NOT 'machine vision' application? They actually are! You think all these are done by a human sitting in front of monitors? No. They are done by advanced AI algorithms for big data analysis. All these are done machines! So Surveillance cameras, feature extraction, face recognition are machine vision......
DeleteAccording to that definition - what is not machine vision? Also smartphone images get processed by AI. So wow, I massively underestimated the size of machine vision ;-) I dont think it is useful to mix all the market spaces but on the other hand does not matter. Surveillance is surveillance no matter if images get processed by AI or manually. The camera itself is for surveillance, so I count it to the surveillance market and not machine vision.
DeleteI don't think that telecentric use should be considered as a standard in industrial machine vision, i refer to telecentric as nich. so if CRA is just a limit for telecentric applications, then this not clasifie the detector as not MV valid
Deleteplease have a look at http://image-sensors-world.blogspot.com/2021/01/yole-on-machine-vision-market.html
Deleteimage/slide 3, title "machine vision cameras". This is what I consider machine vision cameras. It seems Yole shares my view in usage of this term.
If you look at image/slide 2, there is a breakdown into sub categories of 'machine vision' market. Maybe I mixed 'factory automation' with 'machine vision'. My view is on 'factory automation'. This is by far the largest sub segment and I dont know a single camera using a smartsens sensor in factory automation. Not even the chinese players like Hikrobot, dahua etc. seem to use them.