I have a site where the client has recently installed a couple hundred streams of analog WD1 / 960h cameras. For this particular site we would normally encode the streams using Axis encoders. Has anyone had experience encoding these? If so, I have a few questions:
It sounds like the big league manufacturers will be forced to support this on encoders. The lower end/more commonly installed analog manufacturers in our region are predominantly Speco/Pelco/Bosch. It looks like half of Speco's product line is now 960H.
I wouldn't count on that. Last year I asked a number of manufacturers, including Axis, Sony, TKH, Bosch, IndigoVision, Pelco, Dallmeier, Geutebruck and Verint if they planned to "upgrade" their encoders to support 960H. Every one of them stated that they have no plans to do so.
I think the handwriting is on the wall for all things analog. Encoder manufacturers, including VMS companies who manufacture their own encoders, have basically given up on encoder development. Their concentration is on IP almost exclusively.
I also have to wonder at the likelihood we'll ever see ONVIF-compliant encoders for HD-SDI, HDCVI, HD-TVI or AHD. Perhaps some manufacturer of those products will release their own encoder but it appears far more likely that those technologies will be relegated to DVRs and, perhaps, the occasional PCI capture card.
We've tested a few 960H cameras and our conclusion is basically the same as IPVM's: 960H, even when optimized with suitable end devices (essentially DVRs), offers little perceivable advantage over SD (
In fact, the absence of encoders makes 960H's miniscule resolution advantage over SD cameras disappear. When fed into a 4CIF/4SIF/D1 encoder, all analog SD cameras deliver essentially the same resolution; whether their "claimed" resolution is 470TVL, 520TVL, 540TVL, 600TVL, 700TVL or even 960H.
@patrickvonplaten where might we find an updated usage example for the wav2vec robust model that works despite not having the tokenizer_config.json and other files that exist within, say, the 960h model? And could I perhaps help by updating the example on the Hub somehow?
@patrickvonplaten the blog post is nice but is there something with tensorflow? somebody suggested this notebook GitHub - vasudevgupta7/gsoc-wav2vec2: GSoC'2021 TensorFlow implementation of Wav2Vec2 but it is not an official notebook from huggingface. What do you think?
Beware of what you are getting Some organizations sell systems with 960H cameras included and call them 960H solutions, however to take advantage of the superior image quality you must use a 960h DVR which are not readily available in the market presently. Also beware of DVR models that have 960 in the model name because this often refers to the total frames the DVR can support, not the resolution.
You will need to upgrade your DVR/NVR recording device to utilize 960h. You will also require more hard disk storage to accommodate for the larger image size. Although 960h will give you a better image quality over a standard analogue system, the image quality is still superior on an IP or HD-SDI based system. There are a lot of fakes out there, claiming to be 960H, so be careful.
7fc3f7cf58