Maybe ONIV is having some issues with feature implementation, I don't know. But Dahua no doubt has the ability to develop an open source non locked alternative or simply to open up their API protocol. Provided the data is transmitted over poe and provided they allow other software / NVR developers to use their standard for free, their cameras could retain much more universal appeal. They're still not giving their in camera tech away for free as the features themselves are still surely the product of hardware / firmware within the camera (otherwise if in the NVR alone, they could be brought to any camera of any age), and these systems can no doubt be patented / protected. Surely all they need to give away is access to the data stream for 3rd party developers making that available over POE as an open standard or via an openly available API.
As for additional ONIV features in BI, talk to Ken. I don't know what his motives or abilities are for inclusion in BI.
I do believe Dahua are dreaming if they think everyone is suddenly going to buy into a locked eco system. For sure, many large businesses might. However, for those hundreds of thousands of home or small business users who can't afford Dahua's software and don't like the restrictions / cost of NVR's, that isn't going to be an option. Some may continue to use Dahua. It's just buying habits may change from buying for new features to buying upon failure only, reducing the frequency and numbers of new camera purchases.
I also see the opportunity here for another camera manufacturer startup. If Dahua fail to keep their cameras open to all, and Hik follow suit, the market will be missing an open source camera at the more commercial but reasonably priced end, and given the gap in PQ between action cams, phones, DSLRs (some of which is no doubt down to inherent limitations between the platforms eg bit rate vs space, but some of which may not ), I can't help but feel there's a gap in the market opening up for an open source camera with much higher Picture Quality that works on an open standard and sells for around the same price or less than the current big players devices. Maybe there's a new direction for someone like GoPro here as the action camera market becomes crowded out with low price competition.
I see I've been misunderstood here. And unfortunately, I must now defend Dahua. Dahua isn't locking anything down. The market is simply such that only they implement their own functions in their own solutions.
1. ONVIF as a standard and organization is a dead body. In theory, they're developing something, but in practice, they're years behind what individual manufacturers implement on their cameras and NVRs.
2.
Dahua implements four ONVIF standards on cameras and NVRs, providing access to video streams, video storage, full camera configuration, and retrieval/search metadata/events. It's really hard to fault Dahua here. The problem lies more with the VMS, which primarily implement access to video streams and not much else using ONVIF. You could ask the author of BI why they don't implement full metadata from ONVIF M (where there's detection of faces, people, cars, and non-motorized vehicles – like in VideoMetaData/AcuPick)? Or full camera configuration from BI using ONVIF?
3. Dahua, like other manufacturers, develops its own private APIs, where all functions are implemented. After signing an NDA with Dahua,
you can access these private APIs/SDKs (some of them can even be downloaded from the Dahua website without an NDA). Some large VMSs have written additional drivers for Dahua based on these SDKs/APIs.
4. The most valuable for us (and many other open source projects) is the Dahua HTTP API—several hundred functions described in 800 PDF pages - available on the Andy store website. All open source Dahua plugins for Home Assistant, Home Bridge, or Frigate projects - which implement Dahua features (AI, events, intercom/2-way backend support) use this API.
New functionality, such as AcuPick and WizPick, due to camera power constraints, has their processing split between the cameras (which detect objects, partially describe them, and report them to the NVR) and the Dahua NVR/IVSS/DSS (which re-analyzes these objects and builds a 3D model and more detailed descriptions).
This isn't Dahua's invention -
the entire market is moving in this direction. Due to the power limitations of SOCs, cameras only have simple AI onboard. Large AI models require large AI processors, which consume significant power—a modern NVIDIA AI accelerator can consume 300-500 Watts. Therefore, large AI will only be available on centralized Dahua NVRs or Dahua IVSS/DSS Pro servers.
Currently,
AcuPick data is available via the Dahua API. Some data is in ONVIF M, while the rest is in a private SDK/API. The problem is that
no alternative application or VMS has implemented this feature.
Looking at the quality of all popular mobile camera viewing apps that use ONVIF, the reality is unfortunately dismal.
They implement (like most VMS / BI) ONVIF basics for video access and nothing more. They are primarily aimed (like BI) at users of popular, inexpensive consumer cameras and focus on their operation. Support for advanced AI features represents a very small portion of their user base.
Therefore, I'll reiterate: if we want to use Dahua's new features, we're doomed to stick with Dahua's eco-system (Dahua NVR, Dahua apps).
But this isn't Dahua's fault – these solutions are open (accessible via API), but rather the
state of the CCTV / ONVIF app market, which has stuck with
minimal ONVIF support (access to video streams) and doesn't implement anything more.
I repeat again: ONVIF is a dead-end.. And this is a problem, I never saw any interesting ONVIF app with many nice working features.
