Ocelot 2023
Ranked 16th globally in the Validation set
DDP Thesis | Prof. Amit Sethi, Dept. of Electrical Engineering, IIT Bombay
- Utilized various methods for cell detection and classification, including YoloV8 object detection and cell segmentation techniques on the OCELOT dataset consisting of small and large Field-of-View patches from WSIs
- Developed a unified model with DeepLabV3 architecture for cell and tissue segmentation, leveraging the tissue segmentation model’s Large Field-of-View predictions to enhance cell detection and classification
- Attained a F1-score of 0.67 outperforming the author’s baseline of 0.65 F1-score on an undisclosed validation dataset during the Ocelot 2023 Challenge, securing a global ranking of 16th place