WiFi Sensing Kit Teaches ESP32 Movement Recognition

This is a compelling, reproducible edge-sensing experiment rather than a general-purpose “Wi-Fi vision” breakthrough.

• –Cross-board CSI links improve room coverage beyond router-only sensing.
• –A camera-free k-nearest-neighbor model predicts movements after training, while webcam video remains local.
• –The reported 79.5% score beats the 12.5% random baseline, but comes from one person in one living room.
• –Models are highly setup-dependent: moving boards, furniture, or adding people requires retraining.
• –It detects coarse movement classes, not skeletons, precise coordinates, or reliably transferable behavior.