Sementic-Segmentation on LIVECell Dataset

---

LIVECell Dataset (https://sartorius-research.github.io/LIVECell/)

LIVECell is a large, high-quality, manually annotated and expert-validated dataset of phase-contrast images, consisting of over 1.6 million cells from a diverse set of cell morphologies and culture densities.

---

Model used : CANet (Chained Context Aggregation Network) with few modifications.

• Loss Used : dice_loss + (1 * focal_loss)
• Total Epochs : 10
• Training IOU Score : 0.6533
• Validation IOU Score : 0.7483
• Training F1 Score : 0.7812
• Validation F1 Score : 0.8532

---

Model Performance.

---

Visualization of model prediction.

---

Reference :

1. https://www.nature.com/articles/s41592-021-01249-6
2. https://www.sciencedirect.com/science/article/abs/pii/S0262885621002146