Publish gen3 (current) weights + standardized card

README.md

@@ -0,0 +1,110 @@
+---
+license: agpl-3.0
+library_name: ultralytics
+pipeline_tag: object-detection
+base_model: Ultralytics/YOLO11
+tags:
+- object-detection
+- ultralytics
+- yolo11
+- onnx
+- onnxruntime
+- playing-cards
+- card-detection
+- casino
+- live-game-defender
+model-index:
+- name: lgd-cards-gen3
+  results:
+  - task:
+      type: object-detection
+    dataset:
+      type: lgd-poc-video-holdout
+      name: LGD frozen PoC table-video holdout (225 frames, private)
+    metrics:
+    - type: recall
+      value: 0.847
+      name: Pip recall @50 (code-aware)
+    - type: precision
+      value: 0.771
+      name: Precision proxy
+---
+
+# LGD Cards — Gen 3 (day-2 video) · YOLO11s playing-card detector, 52 classes ✅ CURRENT
+
+The **current** generation of the **Live Game Defender (LGD)** playing-card detector, and the model
+LGD serves in production. Gen 2 further fine-tuned on a **second day of real casino-table video**
+(plus a human-QA correction pass), lifting frozen-holdout recall to **0.85**. Locates every playing
+card in a frame and names it by **rank + suit** (52 classes); an empty frame yields no detection
+(built-in "is there a card?" gate).
+
+## Generations
+
+| Gen | Repo | Trained on | Frozen real-video holdout recall | Status |
+|---|---|---|---|---|
+| 1 | [lgd-cards-gen1](https://huggingface.co/sroot/lgd-cards-gen1) | Roboflow `ow27d` v4 dataset | — (dataset-val only) | superseded |
+| 2 | [lgd-cards-gen2](https://huggingface.co/sroot/lgd-cards-gen2) | + day-1 PoC table video | 0.68 | superseded |
+| 3 | **lgd-cards-gen3** (this) | + day-2 PoC table video | **0.85** | ✅ current |
+
+Chip detectors: [lgd-chips-gen1](https://huggingface.co/sroot/lgd-chips-gen1) · [lgd-chips-gen2](https://huggingface.co/sroot/lgd-chips-gen2).
+
+## Classes (52)
+
+```
+10C 10D 10H 10S 2C 2D 2H 2S 3C 3D 3H 3S 4C 4D 4H 4S 5C 5D 5H 5S
+6C 6D 6H 6S 7C 7D 7H 7S 8C 8D 8H 8S 9C 9D 9H 9S
+AC AD AH AS JC JD JH JS KC KD KH KS QC QD QH QS
+```
+
+`C`=Clubs, `D`=Diamonds, `H`=Hearts, `S`=Spades. Full order is in `model.classes.json`.
+
+## Files
+
+- `model.onnx` — ONNX export (run with `onnxruntime`); this is the exact file LGD serves.
+- `model.classes.json` — ordered class-name sidecar (index → card code).
+- `metrics.json` — training config + frozen-holdout verdict vs. the previous generation.
+
+## Training
+
+- **Base:** Ultralytics `yolo11s.pt` (COCO-pretrained), `imgsz=640` (resumed once from epoch 15).
+- **Data:** Roboflow `ow27d` base **+ day-1 PoC video + day-2 PoC video**, auto-labeled (detector
+  proposes pip boxes; an LLM verifies/names over a closed 52-code vocabulary) and materialized as
+  serve-matching tiles — **7,526 train / 1,103 val tiles**, with **14 human-QA label corrections**.
+  Labeling cost **$8.13**.
+- **Hardware:** NVIDIA RTX 3060 (12 GB).
+
+## Metrics — frozen real-video holdout
+
+Scored on a **frozen 225-frame holdout** of our own PoC recordings (never trained on), serve-style
+auto-grid tiling, threshold 50, code-aware. Both generations scored in one run:
+
+| Model | Recall | Precision proxy |
+|---|---|---|
+| **gen 3 (this)** | **0.847** | **0.771** |
+| gen 2 | 0.745 | 0.759 |
+
+Generation-over-generation on the frozen holdout: `stock 0.539 → gen1 0.680 → gen3 0.847`.
+
+> ⚠️ **Not casino accuracy (rule of the project).** The holdout is our own proof-of-concept footage
+> with LLM-verified (not fully human-verified) ground truth. These numbers show the detector
+> *mechanism* and its improvement across generations — not a validated real-world full-deck accuracy
+> claim on a live casino floor.
+
+## Usage
+
+```python
+import json, onnxruntime as ort
+sess = ort.InferenceSession("model.onnx", providers=["CPUExecutionProvider"])
+names = json.load(open("model.classes.json"))   # index -> "AS", "10H", ...
+# 640x640 letterboxed input. The detector boxes corner PIPS (~2 per card): cluster same-code pips
+# within a few box-diagonals into ONE card-level detection before consuming results.
+```
+
+## License & provenance
+
+**AGPL-3.0**, a fine-tune of **Ultralytics YOLO11** (`yolo11s.pt`, AGPL-3.0) — these weights inherit
+AGPL-3.0 and **are not an original work of ours**. Networked deployment triggers AGPL §13 (offer the
+Corresponding Source). `onnxruntime` (MIT) keeps the inference code AGPL-free; the weights stay AGPL.
+
+Built for **Live Game Defender (LGD)** — an on-prem AI integrity monitor for live casino table
+games. © 2026 TechTools s.r.o.

metrics.json

@@ -0,0 +1,29 @@
+{
+ "date": "2026-07-09",
+ "run": "video-campaign-2",
+ "resumed_from_epoch": 15,
+ "data": {
+  "train_tiles": 7526,
+  "val_tiles": 1103,
+  "sources": [
+   "data/video_real",
+   "data/video_real_day2"
+  ],
+  "labeling_cost_usd": 8.13,
+  "human_corrections": 14
+ },
+ "test_eval": {
+  "holdout": "data/video_test (225 frames, 4317 boxes)",
+  "candidate": {
+   "recall": 0.847,
+   "precision_proxy": 0.771
+  },
+  "serving_video_campaign_1": {
+   "recall": 0.745,
+   "precision_proxy": 0.759
+  },
+  "beats_reference": true
+ },
+ "activated": true,
+ "activated_at": "2026-07-09T10:12"
+}

model.classes.json

@@ -0,0 +1 @@
+["10C", "10D", "10H", "10S", "2C", "2D", "2H", "2S", "3C", "3D", "3H", "3S", "4C", "4D", "4H", "4S", "5C", "5D", "5H", "5S", "6C", "6D", "6H", "6S", "7C", "7D", "7H", "7S", "8C", "8D", "8H", "8S", "9C", "9D", "9H", "9S", "AC", "AD", "AH", "AS", "JC", "JD", "JH", "JS", "KC", "KD", "KH", "KS", "QC", "QD", "QH", "QS"]

model.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8b767cdfed2c8e954a9134013ac3d2f2c53be048768d559675be01277a8a8fd1
+size 38233687