Card: mark the earlier 100% presence-accuracy figure as retracted/unverified (per related card)

Corrects three places that still presented the retracted '100% presence accuracy' as fact and adds a correction notice linking to ruvnet/wifi-densepose-pretrained. No new claims added.

README.md

@@ -23,6 +23,9 @@ pipeline_tag: other
 
 # RuView — WiFi Sensing Models
 
+> **Correction notice.** An earlier version of this card reported "100% presence accuracy". The related model card [ruvnet/wifi-densepose-pretrained](https://huggingface.co/ruvnet/wifi-densepose-pretrained) retracts that figure: it says the number was measured on a single-class recording and "says nothing about generalization". Treat any presence-accuracy figure for these models as unverified until a multi-class evaluation is published.
+
+
 **Turn WiFi signals into spatial intelligence.** Detect people, measure breathing and heart rate, estimate pose, track movement, and monitor rooms — through walls, in the dark, with no cameras. Just radio physics.
 
 ## What This Does
@@ -32,7 +35,7 @@ WiFi signals bounce off people. When someone breathes, their chest moves the air
 | What it senses | How well | Without |
 |----------------|----------|---------|
 | **Pose estimation** | **92.9% PCK@20** (17 COCO keypoints) | No camera at deployment |
-| **Is someone there?** | 100% accuracy | No camera needed |
+| **Is someone there?** | presence detection (earlier "100%" figure retracted in the related card; see correction notice) | No camera needed |
 | **Are they moving?** | Detects typing vs walking vs standing | No wearable needed |
 | **Breathing rate** | 6-30 BPM, contactless | No chest strap |
 | **Heart rate** | 40-120 BPM, through clothes | No smartwatch |
@@ -46,7 +49,7 @@ WiFi signals bounce off people. When someone breathes, their chest moves the air
 | Metric | Result | Context |
 |--------|--------|---------|
 | **Pose PCK@20** | **92.9%** | Camera-supervised WiFlow model (v0.7.0) |
-| **Presence accuracy** | **100%** | Never misses, never false alarms |
+| **Presence accuracy** | *retracted — treat as unverified* | The earlier "100%" is retracted in the related card, where it was found to be measured on a single-class recording; it does not show generalization |
 | **Inference speed** | **0.008 ms** | 125,000x faster than real-time |
 | **Throughput** | **164,183 emb/sec** | One laptop handles 1,600+ sensors |
 | **Model size (pose)** | **974 KB** | WiFlow lite, 189K params |
@@ -76,7 +79,7 @@ Trained on real ESP32 CSI (7,000 frames) + real webcam keypoints via MediaPipe (
 | `model-q4.bin` | 8 KB | **Recommended** — 4-bit quantized, 8x compression |
 | `model-q2.bin` | 4 KB | Ultra-compact for ESP32 edge inference |
 | `model-q8.bin` | 16 KB | High quality 8-bit |
-| `presence-head.json` | 2.6 KB | Presence detection head (100% accuracy) |
+| `presence-head.json` | 2.6 KB | Presence detection head (earlier 100% figure retracted; see correction notice) |
 | `node-1.json` | 21 KB | LoRA adapter for room/node 1 |
 | `node-2.json` | 21 KB | LoRA adapter for room/node 2 |
 | `config.json` | 586 B | Model configuration |