Ten years ago, it was enough for cannabis companies to maintain a basic surveillance infrastructure, but with most states now allowing either medical or recreational use of the Schedule III drug, that approach has become harder to sustain, says Kevin Brown, business development director at Hanwha Vision America.
“Each state rolls out their own program [so] more people are worried about standardization across their facilities,” Brown told Facilities Dive in an interview. “They’re thinking about long-term plans for security: what that looks like, how they can be more operationally efficient as opposed to just checking the box for compliance.”
Surveillance requirements are strict in every state, he says. “They’re trying to avoid any kind of diversion of product,” said Brown, whose company manufactures cameras and provides the AI-assisted software that goes with them. “They want to make sure none of the legally grown cannabis product makes its way outside of the legal market.”
That means every inch of a company’s operation must be covered by a camera, he said.
“There shouldn’t be an area in the facility where you can take cannabis and not see it on camera,” he said. “Even if it’s a hallway you’re using to transfer from a growing area to a cure room, that movement through the facility needs to be on camera.”
A large, vertically integrated facility might have more than 500 cameras to cover its operations, from cultivation to extraction to packaging to sales, he said.
“A lot of dispensaries might have 30 to 50 cameras just for the retail space,” he said.
The cameras can generally be the same as the ones used in other industries as long as they meet the requirements states set, he said. “It comes down to megapixels and capabilities like infrared — things like that,” he said. “So most of the cameras we’re selling to a dispensary operator would be the same camera technology that would be going to a typical retailer.”
The one exception are the cameras deployed in the extraction area of facilities, which can be a separate building, room or booth.
Extraction involves the use of volatile gases like butane, carbon dioxide or hydrocarbons to leach out the tetrahydrocannabinol, or THC, from the plant to create the oils, resins and other types of concentrates that go into vape pens, gummies and other products.
States require the cameras to be made of heavy-duty stainless-steel or aluminum housings or be deployed in encasements to prevent a short or other incident from igniting the gases.
“Everything that goes in there needs to be able to safely operate and not cause an explosion,” he said. “It’s basically a camera lens inside of [the enclosure] so nothing escapes.”
Brown said the difficulty of managing different camera systems in different states is pushing facility managers to take a more centralized approach to their surveillance infrastructure.
“Most of them want one pane of glass to look through,” he said. “They don’t want stores in one state using one video management software and stores in another state using a different one, so there’s more long-term planning. That’s a lot of the conversations we're having right now.”
The addition of AI is a game changer for facilities that have hundreds of cameras to monitor because it can be programmed to do the work of people trying to stay alert to red flags, he said.
“Nobody has time to be watching every single camera that’s there,” he said. “So they’re using AI to prioritize the things that matter to them. It could be somebody loitering on the property, entering a restricted area like a vault room or a vehicle pulling into the parking lot after hours.”
The system can also be programmed on how it responds when an alert is triggered. “Is it contacting someone in your operations center? Is it talking [to an unauthorized person] through a speaker?”
Large operations might have an in-house security team that gets the alerts and manages the response; other operations might contract with a third-party for that. “You can set it up so that when your AI detects something and sends out an alert, it goes to a central station, and they can live-talk down through a speaker or take whatever the appropriate action is,” he said.
The AI is also proving to be helpful after there’s been an incident, because it can make the follow-up investigation easier to complete, according to Brown.
“In the past, you would just be manually viewing through lots of footage, lots of cameras, and now with AI on the edge in these cameras, you can search for a specific vehicle or license plate, or somebody wearing a hat and holding a bag and wearing a blue shirt,” he said. “So it makes things a lot more efficient.”
