
Image sensor plus local processing makes AI sensor
The SmartSensor AI image sensor chip platform combines artificial intelligence algorithms with advanced sensors to perform local machine learning.
The company has not provided any details of the approach but said it has selected TSMC’s 3D chip manufacturing process for its “SmartSensor Platform.” This suggests the company is combining a conventional image sensor with a local machine learning processor and possibly additional memory in a multi-die module.
“Sensor-side computing, as its name implies, runs computing functions on the sensor, utilizing the computing power on the chip,” said SmartSens in a statement. The company then listed the usual benefits including: faster data processing speeds, higher system efficiency, greater privacy protection, reduced latency and power saving.
“Sensor-side computing and ‘smart sensor chips’ are one of the major trends in the future development of the IoT industry,” said Tommy Wang, vice president of systems and algorithms at SmartSens, in the statement.
The company did provide examples of smart images captured and annotated using the SmartSensor platform at 4K image resolution. It said the SmartSensor platform could capture and abstract every human face from a complex crowd scene in real-time.
Related links and articles:
News articles:
4Mpixel sensor performs in low light
Sony, ST increase image sensor market share
Analog, MEMS and sensor startups to follow in 2019
