Upload folder using huggingface_hub
Browse files- MASTER_POOL_MODEL.tflite +3 -0
- MASTER_POOL_MODEL.tflite.meta +8 -0
- README.md +34 -0
- metadata.yaml +25 -0
MASTER_POOL_MODEL.tflite
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:99c2d2fbf42d4fb4095ccb94f41f0b1f0b6a56d230d79b9557080587fa1f5c7f
|
| 3 |
+
size 24971937
|
MASTER_POOL_MODEL.tflite.meta
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
MODEL_0_OFFSET=0
|
| 2 |
+
MODEL_0_LENGTH=6242868
|
| 3 |
+
MODEL_1_OFFSET=6242868
|
| 4 |
+
MODEL_1_LENGTH=6242869
|
| 5 |
+
MODEL_2_OFFSET=12485737
|
| 6 |
+
MODEL_2_LENGTH=6243331
|
| 7 |
+
MODEL_3_OFFSET=18729068
|
| 8 |
+
MODEL_3_LENGTH=6242869
|
README.md
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: agpl-3.0
|
| 3 |
+
library_name: tf-lite
|
| 4 |
+
tags:
|
| 5 |
+
- object-detection
|
| 6 |
+
- yolov8
|
| 7 |
+
- billiards
|
| 8 |
+
- tflite
|
| 9 |
+
- cuedetat
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# CueDetat MASTER_POOL_MODEL — Pool Table / Ball Detector
|
| 13 |
+
|
| 14 |
+
The TFLite detector used by the CueDetat Android app for pocket/table and ball/cue detection.
|
| 15 |
+
YOLOv8n, 640×640 input, FP16, NMS in-graph.
|
| 16 |
+
|
| 17 |
+
## Files
|
| 18 |
+
- `MASTER_POOL_MODEL.tflite` — a single binary that **concatenates four sub-models**.
|
| 19 |
+
- `MASTER_POOL_MODEL.tflite.meta` — byte offsets/lengths for each segment (required to load):
|
| 20 |
+
`MODEL_i_OFFSET` / `MODEL_i_LENGTH`. The app maps each segment to its own interpreter.
|
| 21 |
+
- `metadata.yaml` — Ultralytics export metadata (classes, imgsz, stride).
|
| 22 |
+
|
| 23 |
+
## Segments the app actually runs
|
| 24 |
+
- Segment 0 — pocket/table detector. Output: `[1, N, 6]` rows `(y1, x1, y2, x2, score, classId)`.
|
| 25 |
+
- Segment 2 — pool ball / cue detector. Same row layout.
|
| 26 |
+
- (Segments 1 and 3 exist in the file but are unused by the app.)
|
| 27 |
+
|
| 28 |
+
## Classes (from `metadata.yaml`)
|
| 29 |
+
`0: pool-table 1: pool-table-hole 2: pool-table-side`
|
| 30 |
+
(plus ball/cue classes used by the pool head: ball, cue).
|
| 31 |
+
|
| 32 |
+
## Notes
|
| 33 |
+
- Input image normalized to [0,1], RGB, 640×640.
|
| 34 |
+
- License: AGPL-3.0 (Ultralytics-derived).
|
metadata.yaml
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
description: Ultralytics best model trained on /kaggle/working/combined_dataset/data.yaml
|
| 2 |
+
author: Ultralytics
|
| 3 |
+
date: '2026-05-13T07:47:26.724447'
|
| 4 |
+
version: 8.4.49
|
| 5 |
+
license: AGPL-3.0 License (https://ultralytics.com/license)
|
| 6 |
+
docs: https://docs.ultralytics.com
|
| 7 |
+
stride: 32
|
| 8 |
+
task: detect
|
| 9 |
+
batch: 1
|
| 10 |
+
imgsz:
|
| 11 |
+
- 640
|
| 12 |
+
- 640
|
| 13 |
+
names:
|
| 14 |
+
0: pool-table
|
| 15 |
+
1: pool-table-hole
|
| 16 |
+
2: pool-table-side
|
| 17 |
+
args:
|
| 18 |
+
data: null
|
| 19 |
+
batch: 1
|
| 20 |
+
fraction: 1.0
|
| 21 |
+
half: true
|
| 22 |
+
int8: false
|
| 23 |
+
nms: true
|
| 24 |
+
channels: 3
|
| 25 |
+
end2end: false
|