NotSquat helps gym operators see what their existing cameras cannot: occupancy, equipment utilization, dwell time, and congestion.
These member-facing experiences are part of the long-term vision. They are deliberately deferred while we validate the operator analytics product.
Real-time skeletal joint keypoints, knee angle (89°), hip hinge depth (102°), and bar velocity loss (0.52 m/s).
Physical lifts level up your pixel avatar, battle gym-wide raid bosses (Iron Behemoth), and unlock rare gear.
Automated 1RM estimation (Squat 425 lb, Bench 315 lb, Deadlift 515 lb), muscle balance radar, and 98.8% safe form index.
The first beachhead is gym operations. Over time, the same privacy-first video-to-metadata engine could support transit, retail, industrial, and other physical environments.
Current beachhead: occupancy, utilization, dwell time, congestion, and operator reporting. Biomechanics and member progression remain future directions.
High-density private and public hubs. Extracting queue wait times, foot-traffic bottlenecks, dwell curves, and crowd flow without storing invasive personal biometric identity.
The definitive metadata query engine for real-world space. Enabling enterprise operators, insurers, and industrial facilities to query physical activity with the speed and privacy of a database.
We are testing whether existing camera infrastructure can produce useful facility reports without retaining raw video by default. Pilot scope, processing, retention, and any model-training use are defined explicitly with each operator.
A focused path from existing video infrastructure to occupancy, utilization, and congestion reports.
Craft your pixel lifter. Unlock belts, wraps, singlets, and legendary auras.
Physical tonnage damages the raid boss. Work together with your local gym community.
We convert existing CCTV streams into institutional spatial intelligence. In exchange for de-identified CV training vectors, gym owners receive a complete $0/mo data warehouse suite.
2D architectural floor plan heat clustering, wait queues, and crowd density bottlenecks across all equipment zones.
Granular dwell breakdowns (Active work vs Rest vs Phone idle vs Sanitation) with automated rest-nudge queues.
Future direction: movement and safety analysis may follow the operator analytics pilot after technical and legal validation.
Weekly occupancy and congestion reporting to help operators test layout, staffing, and equipment decisions. Retention impact remains to be measured.
5 of 6 racks occupied. Average set cadence: 0.52 m/s. Barbell collision and valgus warnings are normal.
| TIMESTAMP | OPTICAL RE-ID HASH | MEMBER / AVATAR | CAMERA STATION | EXERCISE | LOAD | REPS | BAR VELOCITY | FORM SCORE | SAFETY STATUS |
|---|
Explore a business-case hypothesis. These estimates are not validated results and should be replaced with pilot evidence.
Real-time coordinate normalization, joint velocity formulas, and non-intrusive metadata extraction.
17-keypoint normalized 2D/3D skeleton extraction running at 60 FPS per RTSP stream with zero GPU bottleneck on NVIDIA Jetson edge nodes.
Real-time knee ($89^\circ$), hip ($102^\circ$), and spinal curvature computation with automated 1RM velocity-loss profiling and safety flags.
Our pilot is designed around explicit processing, retention, and data-rights terms. Privacy architecture is a product requirement; compliance depends on the final deployment, controls, and legal review.
Explore the technical blueprints, data schemas, and gamification specifications.
Comprehensive 11-module technical manual & operational blueprint.
Interactive 4-pillar spatial telemetry, dwell times, and actuarial ledger.
iOS & Android store listing screens, kinematics, and avatar RPG studio.
Optical Re-ID, coordinate normalization, and joint kinematics formulas.
Production star schema, fact telemetry, and VBT 1RM profiling pipelines.
16-bit pixel avatar engine, leveling formulas, and raid boss mechanics.
A working visual direction for NotSquat: black and white first, teal and blue only where the system is alive.