IndUSV commited on
Commit
feaf5cc
·
verified ·
1 Parent(s): 111f83e

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +74 -0
README.md ADDED
@@ -0,0 +1,74 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+ ---
3
+ license: mit
4
+ tags:
5
+ - yolov8
6
+ - object-detection
7
+ - computer-vision
8
+ - mobile-phone
9
+ - suitcase
10
+ ---
11
+
12
+ # YOLOv8n Multi-Object Detector (Phone & Suitcase)
13
+
14
+ A fine-tuned **YOLOv8 Nano (YOLOv8n)** model capable of detecting **mobile phones** and **suitcases** in images.
15
+ The model has been trained and validated on custom datasets with diverse backgrounds, lighting conditions,
16
+ and object orientations.
17
+
18
+ ## Model Details
19
+
20
+ - **Base Model**: YOLOv8n (Nano)
21
+ - **Task**: Object Detection
22
+ - **Supported Classes**:
23
+ - `mobile_phone`
24
+ - `suitcase`
25
+ - **Number of Classes**: 2
26
+ - **Input Resolution**: 640 × 640
27
+ - **Framework**: Ultralytics YOLO
28
+ - **Training Date**: 2026-02-06
29
+
30
+ ## Capabilities
31
+
32
+ This model can:
33
+ - Detect **mobile phones** in real-world scenes
34
+ - Detect **suitcases / luggage** in travel and indoor environments
35
+ - Handle multiple objects per image
36
+ - Perform robustly under varied lighting and viewpoints
37
+
38
+ ## Training Details
39
+
40
+ - **Image Size**: 640
41
+ - **Optimizer**: Default YOLOv8 optimizer
42
+ - **Augmentations**: YOLOv8 default augmentations
43
+ - **Device**: NVIDIA GPU
44
+ - **Training Strategy**: Transfer learning from YOLOv8n
45
+
46
+ ## Dataset Information
47
+
48
+ The model was trained using curated datasets containing:
49
+ - Mobile phone images with bounding-box annotations
50
+ - Suitcase / luggage images with bounding-box annotations
51
+ - Real-world indoor and outdoor scenes
52
+ - Multiple object instances per image
53
+
54
+ Dataset annotations follow object-detection standards compatible with YOLO training pipelines.
55
+
56
+ ## Evaluation & Metrics
57
+
58
+ - Quantitative evaluation metrics are available in `results.csv`
59
+ - Qualitative predictions can be seen in validation images from the training run
60
+
61
+ ## Files in This Repository
62
+
63
+ - `Yolov8_SE_2.pt` – Trained YOLOv8 model weights
64
+ - `args.yaml` – Training configuration
65
+ - `results.csv` – Training and validation metrics
66
+
67
+ ## License
68
+
69
+ This model is released under the **MIT License**.
70
+
71
+ ## Acknowledgements
72
+
73
+ - Ultralytics YOLOv8: https://github.com/ultralytics/ultralytics
74
+ - Training datasets sourced and curated for object detection research