They come when they think no one is watching.
But Chula Camps sees them.
The owner of Dogfork Lamp Arts at the end of Quesada Avenue and Fitch Street in San Francisco’s Bayview has witnessed people consistently dump garbage outside her 10,000-square-foot building for the last 10 years.
“It’s frustrating because it doesn’t look good for our business,” she told The Standard “It’s a challenge as it is to get customers to come out to us. We sometimes work late—8, 9, 10 o’clock at night—but for the most part, people feel like it’s an easy dump. We need cameras.”
For decades, District 10 residents and business owners—like Camps—have complained about illegal dumping in their neighborhoods and have been clamoring for more cameras to deter the problem.
According to San Francisco’s Department of Public Works spokesperson Rachel Gordon, the camera surveillance program is still in review.
She said Public Works is finalizing the locations, but the cameras will be installed in “known dumping hot spots,” will capture license plate info and will move from spot to spot.
Yet, in the coming months, residents will get help from an eye in the sky as a pilot program utilizing a drone and artificial intelligence is set to launch into San Francisco skies above the Bayview and Mission hoping to make a difference by increasing the amount of reporting in the neighborhoods.
Brian Johnson, founder of Aerbits—a startup aiming to use unmanned drones and artificial intelligence to streamline city service operations—held a community meeting at the Gratta Wine and Market Thursday evening introducing his drone project to his neighbors.
Johnson, who is a software engineer and Bayview resident, said the inspiration for the program began after he left his position at Zillow in 2021 and had more time to explore the neighborhood with his children. But he’d often have to reroute the meanderings to prevent his kids from running into the trash heaps on their scooters.
“I found that it was really stressful,” Johnson said.
Initially, he would ride his bike around the neighborhood to report the piles of trash. Eventually, he upped the ante by using a drone.
“I thought to myself, ‘I’m an engineer. How can I systemize this?,’” he said. “I went up with the drone, and within two weeks, I had trained a little model with about 30 data points that kind of worked. I’ve spent the last two years working on this.”
Part of the issue, Johnson said, is that 311 data is based on citizen reports.
If someone takes a look at an illegal dumping heat map of San Francisco, Johnson said, areas such as the Mission and Tenderloin look like hotter spots compared to the Bayview. By the same measure of 311 reports, the worst dumping in a commercial corridor is in Bernal Heights.
That didn’t jibe with what Johnson was seeing with his own eyes in his own neighborhood.
“I've been to the Mission, and it's dirty, but, like, we've got a whole other level here—but it’s based on resident reporting,” he said.
The current method of reporting illegal dumping largely involves residents uploading pictures of illegal dumping to the city’s 311 app, which sends them to a call center along with a GPS marker before being dispatched to the Department of Public Works for crews to go clear.
Since it’s driven by complaints and they’re addressed in chronological order, it doesn’t necessarily steer public resources to the places with the greatest need and it doesn’t tend to set crews to the most efficient routes.
Compounding the delays, the city may need to assess what kind of equipment workers should take to a dumping site.
According to Public Works, the goal is for crews to have reported sites cleared within 48 hours of a complaint being filed.
Prior to the drone initiative receiving pilot program status, Johnson said he tested the aerial robot and his narrow artificial intelligence program’s effectiveness by running daily flights, documenting piles of trash through pictures and filing reports to the Public Works while creating his own dataset of what piles look like.
“I’ve labeled 12,000 piles of garbage,” Johnson said.
The drone flies for approximately 35 minutes in a radius of Meade to Revere avenues and from just beyond the west of Third Street to Arelious Walker Drive.
“I've seen the piles get down to like a handful, like five to 10,” he said. “They’re not big piles. I’ve seen them do that by recording daily, and I think we can get down to where some days, they won’t have to do any pickup.”
Once the drone flight is done, it takes another 35 to 45 minutes for the AI to run its process and the reports to be sent.
Johnson said he is optimistic that the Public Works will see the value in the drone pilot program and make it a permanent aspect of the city’s ability to curb illegal dumping. He has also cultivated the program so that images taken from the drone flights will black out houses or vehicles for privacy’s sake.
According to Johnson, other cities and counties in the Bay Area—particularly Oakland, San Jose and Alameda County—have reached out to him about potentially using his drone system to tackle illegal dumping in their jurisdictions.
According to his estimates, Johnson believes between zero and 50 new piles appear almost daily in the Bayview.
The pilot program, he said, will start after the budget is finalized in August and will include two daily flights in both Bayview and Mission for three months.
Even if the drone doesn’t deter future dumping, Johnson said, the data it gleans could inform the city’s allocation of resources.
According to Gordon, if the pilot program is successful, the department will look into putting it out to bid and making it permanent.
“There certainly is potential in this strategy, and we always are looking for innovative ways to combat the scourge of trash on our streets,” she told The Standard in an email.
Camps said she just hopes the eyes in the sky lead to cleaner streets on the ground.
“I’m sure it will develop into a process to educate the residents, but also educate the business owners that we need to do our work to,” she said. "If we see something, we should say something.”
Joel Umanzor can be reached at email@example.com