GHAYTH-LAB/choufli-hall-character-segmentation-tracking-yolov8
Record type:
project
Creator:
GHA
Host:
YOLOv8-based character segmentation for Choufli Hal, a popular Tunisian comedy TV series. This project uses computer vision to detect and segment the series' main characters in video frames, with support for character recognition and tracking
# Choufli Hall Character Segmentation & Tracking YOLOv8
## Overview
This project focuses on detecting, segmenting, and tracking characters from the TV series "Choufli Hall" using a custom-trained YOLOv8 segmentation model. The model was trained on a small dataset and is designed to recognize multiple character classes in video scenes.
It combines:
- custom dataset preparation
- YOLOv8 segmentation training
- object tracking during video inference
- visual overlays for character detection
## Project Goal
The main goal is to detect and follow characters in video footage with segmentation and tracking, making it useful for visual analysis, monitoring, and future improvements in character recognition workflows.
## Version Info
- Version: 1.0
- Model type: YOLOv8 segmentation
- Dataset: small custom dataset
- Training status: prototype / experimental
- Classes: 8
## Tech Stack
- Python
- OpenCV
- Ultralytics YOLOv8
- PyTorch
- Roboflow
- Windows OS
- Computer vision pipeline
## Dataset
The project uses a custom dataset exported from Roboflow and stored in the following folder:
- `choufli hall carcters segmentati.v5-please-work-wtf.yolov8/`
Dataset configuration file:
- `data.yaml`
Detected character classes:
- AZZA
- Dalanda
- beji_matrix
- fouchika
- jannet
- sboui
- slimane
- zeineb
## Model Configuration
The model was trained using a YOLOv8 segmentation architecture with these settings:
- Base model: `yolov8n-seg.pt`
- Epochs: 70
- Image size: 640
- Batch size: 16
- Workers: 0
- Training script: `train.py`
- Best checkpoint: `best.pt`
## Project Structure
```text
choufli-hall-character-segmentation-tracking-yolov8/
├── best.pt
├── yolo26n.pt
├── yolov8n-seg.pt
├── main.py
├── train.py
├── readme.md
├── tempCodeRunnerFile.py
├── Failure image.jpg
├── مسلسل شوفلي حل الموسم 2005 - الحلقة الثالثة.mp4
├── runs/
│ └── segment/
│ ├── train/
│ ├── train-2/
│ ├── train-3/
│ ├── train-4/
│ ├── train-5/
│ ├── train-6/
│ ├── …
Similar
Phonetic Segmentation of the UCLA Phonetics Lab ArchiveGhayth-Bouzayeni/tunisian-voice-pipelineSolar Panel Segmentation in High-Resolution Satellite Imagery: A YOLOv8-GIS Approach in the Marrakech-Safi Region, MoroccoA deep learning-based system for coral reef image segmentation using YOLOv8 with EfficientNet-B0 in Indonesian watersEvaluating AI for Palm Tree Disease Detection: A Comparative Study of YOLOv8 Object Detection and U-Net Segmentation Using UAV ImageryEmmanuel-Acquah/g-yolov8-msv-ghana