With the launch of a higher-end surveillance camera, Dropcam signaled it is broadening its focus to become a platform for connected homes. Unveiled Thursday, Dropcam Pro ($199) offers substantial hardware and software upgrades over its nearly two-year-old predecessor, which the company will continue to offer at $149. Most notably, with the inclusion of a Bluetooth low-energy radio in Dropcam Pro, the company has opened up the potential for the camera to communicate with and receive cues from other sensors.
Dropcam built its brand on simplicity. As a testament to its success, it has become one of the most recognizable names in consumer surveillance cameras. Dropcam (previously Dropcam HD) is consistently one of the three most popular products sold in Amazon's camera and photo department, typically hovering at No. 2 behind GoPro. The San Francisco company is hoping its focus on simplicity will carry over with a connected home platform.
"We look at [Bluetooth low energy] as a bet on the future to be a platform for those connected devices," CEO and cofounder Greg Duffy told Fast Company. The integration of Bluetooth low energy indicates a lot of potential, since that's the protocol used by newer wireless devices. Out of the box, Bluetooth low energy allows for mobile setup, but Duffy noted down the line, it will allow Dropcam Pro to talk to other connected home devices, such as smart pet collars and door sensors.
"No normal person knows what a connected home is," said Duffy, who entertains the possibility of selling other smart home devices. "We're hoping we will get them piece by piece into a connected home. They'll start off with a sensor, a camera, and slowly add sensors to specific things they want to see and sense around their homes."
The dual-band Dropcam Pro packs in an image sensor measuring one-third of an inch, twice the size of the one in its predecessor, giving it better performance in low-light conditions. Its six-element glass lens provides a swooping 130-degree view of a room, a wider view than the 107-degree vantage of the previous model. The app has also been updated so viewers can pan and zoom up to eight times into specific areas, and an enhance button improves the clarity of the enlarged footage. "You lose no quality with enhance," Duffy mentioned. "There's much more resolution in the sensor than it's able to stream, so it does give you better image quality. It's not just blowing up the image."
While the hardware upgrades are substantial, perhaps more interesting is the software. The most glaring issue with Dropcam hasn't been low-light performance or image quality, but motion detection, which has also been Dropcam's most useful feature--it's what catches burglars after the fact (or even in the act) and what landed Dropcam on a 20/20 special on nasty neighbors recently. But the camera has trouble telling significant motion events (such as someone walking through the door) from insignificant ones (a cat walking by or a flicker of light). As a result, Dropcam owners can receive dozens of phone or email notifications throughout the day; over time, they might ignore all of them. (Those who pay for the cloud-recording service have the option to view footage up to 30 days old.)
To solve the problem, Dropcam has a multipronged solution, parts of which have already been implemented (eg. users can choose when to turn on and off alerts) and others under way. With Bluetooth low energy, sensors, such as one for the door, can help Dropcam identify and categorize different events to filter out noise. Furthermore, the company has hired computer vision researchers to incorporate artificial intelligence into the camera. "You as a human can say that's just light going across a wall, or if someone walks in, that's a person walking in," Duffy explained. "To a computer, it's like speaking Greek. Video analysis is still early in the field, so we're able to do things no company has been able to do and give you more information."
With Dropcam Pro, the company is testing activity recognition so that the camera automatically categorizes events, only alerting consumers when certain types of motion are detected. "On the software side, activity recognition is most exciting because it's our first machine-learning feature," Duffy said. Though activity recognition is only a beta feature, he posits that Dropcam Pro's artificial intelligence could get smarter and more specific over time. "Maybe in the future, it can not only determine if it's a pet or person walking through a frame, but which pet or which person," he added.