topic: pose
21 modules sharing the pose name family (derived from the tree's prefix discipline).
The 'pose' topic family in the Nishi sovereign ecosystem focuses on body pose estimation and analysis, integrating neural network models with formal verification. Modules like nx_pose_cnn produce heatmaps for pose detection, while nx_pose_backprop_gate ensures the correctness of multi-layer backpropagation in pose networks. nx_pose_coach evaluates user motion against reference tracks, supporting real-time feedback and scoring, essential for training and validation within the ecosystem.
auto-narrated by the local model from this topic's module headers; links verified against the wiki index.
narrated overview -- maintained by the narration lane, module links verified against this wiki.
| module | description | lines | funcs |
|---|---|---|---|
| nx_pose_backprop_gate.nx | proves the MULTI-LAYER pose-net backprop COMPOSES correctly (the "scale build-2" | 99 | 3 |
| nx_pose_benchmark_audit.nx | CRITIC + ADVERSARY audit (operator 2026-07-14: "make sure we aren't self validating | 78 | 3 |
| nx_pose_classify.nx | classify body pose from face + torso geometry. | 142 | 1 |
| nx_pose_classify_test.nx | 64 | 1 | |
| nx_pose_cnn.nx | T2 POSE FRONT-HALF: the sovereign software-f32 pose-CNN FORWARD PASS -- the heatmap PRODUCER | 52 | 3 |
| nx_pose_cnn_gate.nx | executed proof of the sovereign f32 pose-CNN FORWARD PASS end-to-end into the integer | 106 | 3 |
| nx_pose_coach.nx | the COACH's read (roadmap R8): compare a USER pose track to a REFERENCE track and say | 72 | 7 |
| nx_pose_distill_gate.nx | BUILD-2 core proof: a small STUDENT pose net TRAINS (the distillation loop) to | 107 | 5 |
| nx_pose_elm.nx | the NONLINEAR learner for pose-from-silhouette (roadmap R2 v1). The linear perceptron on raw | 69 | 6 |
| nx_pose_estimate.nx | LEARNED pose localization from an image (roadmap R2 v0, the video->POSE keystone gap). | 75 | 4 |
| nx_pose_estimate_min.nx | minimal driver to isolate whether nx_pose_estimate (lib) compiles. expect_exit:0 | 12 | 1 |
| nx_pose_keypoints.nx | T2 POSE, the WEIGHTS-FREE back half of a heatmap pose estimator (HRNet/OpenPose-class, | 59 | 3 |
| nx_pose_keypoints_gate.nx | gate for the weights-free pose back-half (heatmap -> keypoint). Synthetic heatmaps | 63 | 3 |
| nx_pose_naturalness.nx | Tier 6 biomechanical pose validation. | 355 | 5 |
| nx_pose_preprocess.nx | ViTPose top-down preprocessing: real image + GT bbox -> vitpose_input.bin. | 183 | 7 |
| nx_pose_retarget.nx | RETARGET a pose track (joint world POSITIONS) onto our rig's joint ANGLES (roadmap R3, | 73 | 4 |
| nx_pose_retarget_gate.nx | prove RETARGET (roadmap R3) inverts the FK: known joint ANGLES -> FK -> joint | 102 | 6 |
| nx_pose_score_one.nx | the FIRST real external-GT score: map the ViTPose port's 48x64 heatmap | 69 | 3 |
| nx_pose_skeleton.nx | T2 POSE, the geometric skeleton analysis that turns keypoints (from nx_pose_keypoints) | 52 | 3 |
| nx_pose_skeleton_gate.nx | gate for the skeleton posture geometry (nx_pose_skeleton). Synthetic skeletons | 66 | 3 |
| nx_pose_student_distill.nx | BUILD-2 REAL DISTILLATION: our OWN sovereign student pose net learns to reproduce | 156 | 8 |