Jiaqi Geng, Dong Huang, Fernando De la Torre
2022 · arXiv
Advances in computer vision and machine learning techniques have led to significant development in 2D and 3D human pose estimation from RGB cameras, LiDAR, and radars. However, human pose estimation from images is adversely affected by occlusion and lighting, which are common in many scenarios of interest. Radar and LiDAR technologies, on the other hand, need specialized hardware that is expensive and power-intensive. Furthermore, placing these sensors in non-public areas raises significant privacy concerns. To address these limitations, recent research has explored the use of WiFi antennas (1D sensors) for body segmentation and key-point body detection. This paper further expands on the use of the WiFi signal in combination with deep learning architectures, commonly used in computer vision, to estimate dense human pose correspondence. We developed a deep neural network that maps the phase and amplitude of WiFi signals to UV coordinates within 24 human regions. The results of the study reveal that our model can estimate the dense pose of multiple subjects, with comparable performance to image-based approaches, by utilizing WiFi signals as the only input. This paves the way for low-cost, broadly accessible, and privacy-preserving algorithms for human sensing.
Metadata: arXiv. Abstracts retain their original authors’ rights.
Original HN comments, with labels from simple text rules. A label describes the whole comment and may miss context or irony. Read the discussion to make up your own mind. About the labels
7 selected excerpts · 7 total mentions
Thanks to the beamforming that's central to modern Wi-Fi and 5G cellular standards they can even see through your walls using the backscattered energy of the steered beam, like an airport security scanner for the entire planet: https://arxiv.org/pdf/2301.00250.pdf https://people.csail.mit.edu/fadel/papers/wivi-paper.pdf
I saw "over-engineering" in the title and half expected to see Wi-Fi signals being abused to detect human bodies in the room. https://arxiv.org/abs/2301.00250
It's really interesting that people seem to be getting [0] such good results from wifi imaging while in practice (and my personal experience at least) bluetooth which operates in the same band can't even muster a reliable 2D angle of arrival with a 6 inch array of 5 antennas. I highly doubt any of this will work very well outside lab conditions. [0] https://arxiv.org/pdf/2301.00250
I'm a little skeptical of the paper in question: https://arxiv.org/pdf/2301.00250.pdf If you look at the training and test set info: > We report results on two protocols: (1) Same layout: We train on the training set in all 16 spatial layouts, and test on remaining frames. Following [31], we randomly select 80% of the samples to be our training set, and the rest to be our testing set. The training and testing samples are different in the person’s location and pose, but share the same person’s identities and background. This is a reasonable assumption since the WiFi device is usually installed in a fixed location. (2) Different layout: We train on 15 spatial layouts and…
test on 1 unseen spatial layout. The unseen layout is in the classroom scenarios. Depending on how they selected various frames -- let's just say it was random -- the model could have learned something to effect of "this RF pattern is most similar to these two other readings I'm familiar with" (from the surrounding frames) and can therefore just interpolate between the resulting poses associated with those RF patterns (that the model has compressed/memorized into trained weights). If you look at the meshes between the image ground truth and the paper's results, you'll see that they are strikingly similar. I find this also suspect because WiFi-band RF interacts a lot more with water than with clothes and so you would expect the outline/mesh to get the "meat-bag" parts of you correct but not be able to guess the contours of baggy clothes. That is... unless it has memorized them from the training set.
Related: 2013 - gesture recognition from Wi-Fi https://wisee.cs.washington.edu/ 2013 - see through walls with Wi-Fi https://people.csail.mit.edu/fadel/papers/wivi-paper.pdf 2023 - dense pose estimation from Wi-Fi https://arxiv.org/abs/2301.00250
- FluLM : Using language model to generate, synthesize and evaluate new symptoms and variants in various flu strains. (Something like "Learning the language of viral evolution and escape" https://pubmed.ncbi.nlm.nih.gov/33446556/ ) - PornDiffusion : A tailored to your need addictive auto-generated sexy Chat-bot with videos sucking your time and soul like a Succubus version of "Her". - Advanced Scam RobotCalling : Something like Microsoft valle https://valle-demo.github.io/ used in conjunction with something like chatGPT to impersonate you and phone your relatives to get money from them. - Kompromat collector : As first generation AI model like chat GPT have been dubbed…
"bullshitter", there has been more research into identifying and mine true facts from data. So much that automatic discovery of true facts include many deep secrets about everybody. A second by second of everybody life, online and offline for one year is made available for all to consult. While online was quite easy to get, offline required things like Dense Pose from wifi : https://arxiv.org/abs/2301.00250 and data from SOLI devices combined and index into a Shodan like network, that you can use in natural language to instantly dig dirt on anybody. - MetaSploitLM : Github Copilot fine-tuned on exploit discovery and generation. - BubblePopper : Get out of your filter bubble thanks to custom recommendation engine : Humans filter bubble exists to avoid humans being overloaded by Data they have not yet grasped. Popping the bubble too fast in an addictive way can result in an epidemic of advanced cynicism tearing societies apart, especially once revealed that most conspiracies are true :) - DeepNuke : Once the source code of the universe had been discovered by automatic discovery tools with open data from the LHC, it was realised that it is in fact quite easy to build a Cold Fusion Nuke from household items.
Good shout. I checked the actual article (here: https://arxiv.org/pdf/2301.00250.pdf, link in the article is broken), and seems you would need multiple routers: > [...] three WiFi transmitters and three aligned receivers [...]. It should be noted that many WiFi routers, such as TP-Link AC1750, come with 3 antennas, so our method only requires 2 of these routers. So basically, one on each side of the subject, one to transmit the signal, the other to receive, and then you can work out the interference pattern. It may be possible to do the same with the transmitters and receivers on the same physical device by measuring the reflected signal, but that doesn't seem to be what…
they were testing. It seems that the person needs to be between the transmitter and receiver, so I'm not even sure if triangulation would work, although figure 2 implies that you need the 'line of sight' between the routers to be blocked (three 'transmitter' routers and three 'receiver' routers).
More from the same topics.
Tom B. Brown, Benjamin Mann, Nick Ryder +28
Recent work has demonstrated substantial gains on many NLP tasks and benchmarks by pre-training on a large corpus of text followed by fine-tuning on a specific task.
Chiyuan Zhang, Samy Bengio, Moritz Hardt +2
Despite their massive size, successful deep artificial neural networks can exhibit a remarkably small difference between training and test performance.
Mariusz Bojarski, Davide Del Testa, Daniel Dworakowski +10
We trained a convolutional neural network (CNN) to map raw pixels from a single front-facing camera directly to steering commands.
David Manheim, Scott Garrabrant
There are several distinct failure modes for overoptimization of systems on the basis of metrics.