That's exactly what Tesla is doing with their validation vehicles, the ones with Lidar towers on top. They establish the "ground truth" from Lidar and use that to train and/or test the vision model. Presumably more "test", since they've most often been seen in Robotaxi service expansion areas shortly before fleet deployment.
I don't have a specific source, no. I think it was mentioned in one of their presentation a few years back, that they use various techniques for "ground truth" for vision training, among those was time series (depth change over time should be continuous etc) and iirc also "external" sources for depth data, like LiDAR. And their validation cars equipped with LiDAR towers are definitely being seen everywhere they are rolling out their Robotaxi services.
They've acknowledged an issue now on the status page. For me at least, it's completely down, package installation straight up doesn't work. Thankfully current work project uses a pull-through mirror that allows us to continue working.
And this is why my setup will be using Reolink cameras integrated locally via HomeAssistant and Frigate. Detection runs locally on cameras and/or in Frigate, HA manages events and UI, and the only way to access any of it remotely is via VPN, no "cloud" anything.
If the authorities come knocking with a warrant, or frankly, even a nicely-worded sensible request, sure, have at it. But ain't nobody accessing the footage unnoticed and without my approval.
I bought Reolink PoE cameras because they support standard RTSP and I could put them on their own VLAN where they can't get to the Internet. I can still use the Reolink app to view them when I'm on the LAN, or through Wireguard when I'm not. I use ffmpeg to save streams to a big disk. Works great.
I do the same and it works well. But do you see the Reolink app hammering domains of the form p#p#.reolink.com? I have to kill the app between uses it it drains my phone's battery because of this.
I've never looked at what the Android app is doing on the network, but maybe I should. I usually just keep it open for a few minutes when I'm waiting for a delivery or I hear someone ring the doorbell, so I'm not too worried about it piping lots of video to Reolink HQ. I haven't noticed much of a battery drain having used it earlier in the day.
Yeah, this is my idea as well. Good to know that the Reolink app still works locally and via Wireguard, that means I'll have less UI to set up in HomeAssistant.
My reolink runs off power over Ethernet. Here in the uk, there have been reports of Amazon drivers using wifi deauth-ing devices to kill the ring cameras when they are near your property. My parents recently experienced this.
Absolutely, I will be running PoE wherever possible, definitely for the doorbell. Not just because of Amazon drivers, I've read reports of burglars using WiFi jammers to make sure wireless cameras are useless during a break-in.
Well this is disturbing. I guess my weekend project is running ethernet to my front door.
My problem is that the area around my front door where the doorbell is installed is solid timber, so it's not just a simple ethernet drop. I'm honestly not even sure how the builders ran the existing wire to that location. Maybe my only option is to add a second backup camera in a location where an ethernet install is easier.
Cops will also go after your networking prior to busting down your door. Same applies regardless of a correct and valid warrant so do with that as you will.
I'm sure the cops who trashed Afroman's place would have loved that ability.
I realize that if the authorities target me specifically, there's not much I can do. Even though I am not in the States and do not expect my local police to be quite as... forceful.
However, I do not intend to make it easy to just grab my footage along with any other that can be found, without at least asking.
Yeah, this is more of a "If police screw up, they'll still make it a 'you' problem". IIRC the example I was thinking of was federal agents attempting to threaten an anti-ICE activist.
I've had better experience with Armcrest cameras but same setup. Moved from using Arlo to a completely local setup and do not regret it one bit.
Worst part was just running ethernet to the spots where the cameras needed to go (only crawlspace access) but nice not having to charge batteries and even nicer knowing i'm not sending video to netgear anymore.
If I had of had a webcam on my front door a few weeks ago I would have been able to identify the thieves that broke into my car and stole a bunch of stuff whilst I was asleep.
Since then I have "cammed" up, but I use my own hard wired network and connected to a Pi5 with a Hailo8 chip running frigate.
No third party apps, just the fun of more stuff on the network. I do run a Cloudflare tunnel on the PI so that I can connect to Frigate from anywhere when I get alerts.
But basically, it's me and only me accessing the content of those cameras. However I do plan to configure Frigate to upload the alerts and detections into S3 with a three month lifecycle.
Oops I initially posted this comment at the top level before I saw this thread:
I assume some of the concern around this is that folks don't want to live in a panopticon. If that's your objection, I can't really help with that. On the other hand, if your objection is that you don't want a backdoor built into your video doorbell (even one that you must opt into), I'm happy to report that there are good non-Ring options.
I switched to a Reolink video doorbell, and it has decent support for local-only operation. It has the ability to save footage to a local micro SD card, and if you're worried about someone stealing the entire doorbell and losing your footage, it also supports RTSP (a common IP camera protocol). You can even have it upload footage to a FTP server on a schedule. It also supports PoE if you're lucky enough to have ethernet at your doorbell, or don't mind doing the drop yourself.
Set up does require an app, but you don't need to use the app after that. Push notifs also require egress, but, iiuc, this is mostly because of how push notifications work. Push does NOT require a paid subscription.
I personally just use the app, but it's nice knowing that if Reolink tried to pull a fast one, I could just block egress on my VLAN and use it locally.
If you'd rather just go completely app-less, I imagine a dumb doorbell paired with an IP camera and a local ZoneMinder [1] install would provide most of the benefits of something like Ring. Of course, the tradeoff is you now have a second job being sys admin of your homelab. Pick your poison, I guess.
This won't matter when all of your neighbors have a Ring doorbell, you will still be recorded every time you leave your house. You are just making yourself more noticeable and disagreeable to the surveillance apparatus.
I have 2 Reolink PoE cameras, and they just hopped on the network with DHCP and I used a web browser to configure them. No app needed. I did install the app later, once the cameras were safely on their own VLAN, and I could connect to them by internal host name or IP address.
I use NetNewsWire as a frontend, and self-hosted FreshRSS as backend for sync and feed management. Works a treat across multiple devices, Mac/iOS/iPadOS and web.
Zipline drones fly quite high, and instead of descending and landing to deliver their payload, they hover at altitude and lower a "delivery pod" down on a wire. The pod also has maneuvering capabilities, but all of its thrusters are fully enclosed, and it's designed to not cause any damage even it if collides with something during descent or ascent. Overall, a very clever design that should be safer, create no noise on ground level, and be able to deliver into much smaller and more confined landing zones.
Sure, but it's the launch/landing site that is in question. It's not just a helipad set up. It's very sci-fi looking and looks very complicated. I'm wondering if the drone itself lands at the top and then lowers the pod for loading to keep them out of reach of the employees. Probably even touted as a safety feature. I just haven't seen the system in operation, and their website just ignores this part as it's not something necessary for marketing.
There is a bunch of videos on Youtube on Zipline, some of them from the company itself (this one showing specifically the "platform": https://www.youtube.com/watch?v=airEzThGlx8), and some from various tech people looking into the whole thing (like Markus Brownlee: https://www.youtube.com/watch?v=88yQTzlmsiA). Probably a better overall source of into than their website.
RTO may work as long as your teams are geographically co-located and return to the same office. In my experience, a lot of teams in recent years have been staffed without this aspect in mind, because with remote it made no difference. So now, even with RTO people still have to constantly sit in remote meetings / work rooms with the rest of the team in other office(s), and the benefit of in-person collaboration is still lost. Arguably, this "remote between offices" mode is the worst of them all, because remoting in from the office almost always results in an inferior experience compared to remoting in from a well-tuned home setup.
This is a thing I've been constantly bringing up with our company. I _do_ think that local collaboration in the office is vastly more effective than remotely, _but_ only given that the entire team is co-located.
As soon as there are any remote members involved, the local collaboration benefits are lost, and a mixed team becomes less effective than a fully-remote one - because few offices offer the necessary space and equipment for large groups of people to participate in remote / hybrid meetings and work groups effectively (most / almost all existing conference room equipment is complete junk). Unfortunately, fully co-located teams appear to be a thing of the past, and as you say, mandates aren't going to help here.
Pretty nice. How does it handle multiple displays? I've set it up with local Ollama, and it seems to only record and analyze one of my two secondary displays. It would be ideal if I can select which one is used if the recording is limited to a single display, or even better if it can record and analyze the entire multi-monitor desktop surface.
edit Nvm, it seems it always records the display that is currently in focus. That is probably the better way to handle it, since it automatically solves the "ignore what's shown but not interacted with on secondary displays" problem.
This was my question also, I think "even better if it can record and analyze the entire multi-monitor desktop surface" would be the best option. I don't know what the impact of that would be on both recording size and AI processing time, but just because one monitor is focused doesn't always mean what's happening on another is ignored. Some examples: an ongoing meeting or watching a video on one screen while taking notes on another; or coding on one screen and a browser/app auto-refreshing on another.
Yep, you figured out how it works! That was the easiest solution I could come up with. I'm sure theres additional context on other screens but this was a good 90/10 solution.
I never liked titanium very much, but it's better than aluminum for sure. Now back when they did stainless, that I thought was awesome and looked and felt way more premium.
I get a new iPhone as a work phone bi-yearly, and get to pick the model, with co-pay for anything above the most basic one (currently 16E). In a month or so, it'll be that time again.
Last time I went with the non-Pro option was 6 years ago with the 11. Back then I regretted it, it was a clear downgrade from my previous iPhone X (which I consider the first "Pro" iPhone) - largely because of the screen, 11's LCD really wasn'g great after the X's AMOLED. Since then, I've had the 13 Pro and 15 Pro, both have been great, even though both are a bit weak in the battery department. Now I'm again considering the base 17.
Not interested in the Air or the 17 Pro Max, and the 17 Pro doesn't really seem like it's worth the extra cost to me, since the base model seems really good this time around. At a glance, the things you get with the 17 Pro for extra 350€ (German pricing) are:
- Unibody, but same aluminum material
- Same size, same _exact_ screen (all specs match)
- A19 Pro chip vs A19 - relevant for gaming (I don't) or other highly intensive applications (video editing I guess? again, not something I do a lot of)
- Better camera system - arguably the only relevant thing for a daily driver here
- LiDAR - very limited usefulness, in my experience having it on the 15 Pro
- USB 3 with 10 Gbit speeds vs USB 2 speeds on the USB-C port - relevant if you transfer large files like ProRes video, for me it's basically only when doing local backups to the Mac to install iOS betas, so largely irrelevant
- 31 hours battery life vs 30 - marginal difference
- Same charging, connectivity, everything else really
Will likely end up visiting an Apple Store to fiddle with both options, and look at the co-pay prices that our company will offer, then decide. But so far, it's much less of a clear Pro choice than it was in the previous years.
It's not because your work gives you the opportunity to get a new iPhone that you *need* to get one... you can also downscale the consumerism, the planet is better that way
It's not because the site gives you a reply button that you need to press it... you can also downscale the server & energy usage, the "planet" is better that way
The year of the 11 lineup was the one year where the standard iPhone <LATEST_NUMBER> was a clear downgrade from last year's. It was a refresh of the iPhone XR from the previous year, which was the budget option compared to the flagship XS.
Yes, and the XS was the follow-up to the previous X and thus also the "Pro" model, before they were explicitly named as such. So my move that year was indeed a downgrade. Now I'm kind of worried of doing the same thing again - but spec-wise, the base 17 is really not at all a downgrade from my current 15 Pro, apart from the camera department.
Base iPhone 17 is the first year since the Pros existed where I think most people are not missing out by getting the non-pro.
I'm certainly not a photographer but I got the 16 Pro because of ProMotion. If I were upgrading this year, I'd probably go for the standard 17 because it finally has practically everything a standard user could expect.
Phones default tonusing Through the Lens autofocus. This is closed-loop and pretty fastthese days.
Only when it's too dark work effectively do they switch to lidar/tof open loop autofocus systems.
