Upload folder using huggingface_hub

README.md

@@ -1,7 +1,96 @@
 ---
-license: mit
-language:
-- en
-base_model:
-- Ultralytics/YOLO11
----
+language: en
+tags:
+- pose-estimation
+- yolo
+- sports-analysis
+- biomechanics
+- computer-vision
+license: apache-2.0
+---
+
+# Custom YOLO Pose Estimation for Sprint Analysis
+
+## Overview
+This repository contains a **custom-trained YOLO pose estimation model** designed specifically for **athletic movement analysis**, with a focus on **sprint biomechanics and stride detection**.  
+
+The model extends the **COCO 17-keypoint schema** and applies **temporal smoothing** during inference for slow-motion footage. It is optimized for **single-person side-view videos**.
+
+---
+
+## Model Details
+
+- **Framework:** Ultralytics YOLOv11 Pose  
+- **Model variant:** yolo11n-pose  
+- **Input size:** 640×640  
+- **Training epochs:** 100  
+- **Device:** CPU  
+- **Precision:** FP32  
+- **Pretrained:** Yes  
+
+---
+
+## Key Features
+
+- Side-view sprint video optimized  
+- Slow-motion analysis  
+- Single-person assumption  
+- Temporal smoothing compatible  
+- Exportable to **ONNX** and **TorchScript**  
+
+---
+
+## Performance Metrics
+
+Metrics computed on **side-view slow-motion sprint clips**:
+
+```json
+{
+  "detection_metrics": {
+    "precision": 0.995,
+    "recall": 0.952,
+    "mAP50": 0.979,
+    "mAP50-95": 0.938
+  },
+  "pose_metrics": {
+    "precision": 0.500,
+    "recall": 0.488,
+    "mAP50": 0.493,
+    "mAP50-95": 0.457
+  },
+  "epochs": 100
+}
+```
+
+---
+
+## Usage
+
+```python
+from ultralytics import YOLO
+
+model = YOLO("best.pt")
+results = model.predict("input_video.mp4", conf=0.25, iou=0.7)
+results.show()
+results.save("output_video.mp4")
+```
+
+---
+
+## Citation
+
+```bibtex
+@misc{mehdid2026yolopose,
+  title={Custom YOLO Pose Estimation for Sprint Analysis},
+  author={Mehdid, Samy Abderraouf},
+  year={2026}
+}
+```
+
+---
+
+## License
+
+```mit
+This project is licensed under the MIT License.
+```

best.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05bbb32a09e7e74d334e1149f5ef409f4901b49ba7d64582f6469b0ba73ea0be
+size 11854983

best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:244e6881215cdb0ba251d2100163aaf2f2f4942cf2acf118305a9756a20a7aa3
+size 6048776

config.yaml

@@ -0,0 +1,92 @@
+model:
+  framework: ultralytics-yolo
+  task: pose-estimation
+  variant: yolo11n-pose
+  pretrained: true
+  input_size: [640, 640]
+  device: cpu
+  precision: fp32
+
+keypoints:
+  schema: coco
+  count: 17
+  flip_indices:
+    - [1, 2]
+    - [3, 4]
+    - [5, 6]
+    - [7, 8]
+    - [9, 10]
+    - [11, 12]
+    - [13, 14]
+    - [15, 16]
+
+training:
+  epochs: 100
+  batch_size: 16
+  optimizer: auto
+  initial_lr: 0.01
+  final_lr_factor: 0.01
+  momentum: 0.937
+  weight_decay: 0.0005
+  warmup_epochs: 3
+  loss_weights:
+    box: 7.5
+    pose: 12.0
+    keypoint_objectness: 1.0
+    classification: 0.5
+    dfl: 1.5
+  data_augmentation:
+    mosaic: 1.0
+    mixup: 0.0
+    cutmix: 0.0
+    horizontal_flip_prob: 0.5
+    vertical_flip_prob: 0.0
+    hsv:
+      h: 0.015
+      s: 0.7
+      v: 0.4
+    scale: 0.5
+    translate: 0.1
+    auto_augment: randaugment
+    erasing_prob: 0.4
+  deterministic: true
+  seed: 0
+
+evaluation:
+  validation_split: val
+  metrics:
+    detection:
+      precision: 0.995
+      recall: 0.952
+      mAP50: 0.979
+      mAP50-95: 0.938
+    pose:
+      precision: 0.500
+      recall: 0.488
+      mAP50: 0.493
+      mAP50-95: 0.457
+
+inference:
+  confidence_threshold: 0.25
+  iou_threshold: 0.7
+  max_detections: 300
+  single_person_assumption: true
+  temporal_smoothing: external
+  optimized_for:
+    - side-view footage
+    - slow-motion video
+
+export:
+  supported_formats:
+    - torchscript
+    - onnx
+  simplify_onnx: true
+
+notes:
+  use_case: sprint biomechanics and stride analysis
+  camera_view: side
+  video_type: slow-motion
+  limitations:
+    - Not optimized for frontal or oblique camera views
+    - Performance may degrade with heavy occlusion
+    - Single-athlete scenarios only

metrics.json

@@ -0,0 +1,16 @@
+{
+  "detection_metrics": {
+    "precision": 0.99505,
+    "recall": 0.95229,
+    "mAP50": 0.97915,
+    "mAP50-95": 0.93786
+  },
+  "pose_metrics": {
+    "precision": 0.5,
+    "recall": 0.48808,
+    "mAP50": 0.49316,
+    "mAP50-95": 0.45724
+  },
+  "epochs": 100,
+  "notes": "Metrics computed on side-view slow-motion single-person sprint clips."
+}