Description

Classifier for 5 room types: bedroom, bathroom, dining room, living room, kitchen) from images despite wide variability in room appearances. Applications in real estate, interior design, robotics, and security for automating room identification and improving task efficiency.

Example predictions:

Accuracy analysis:

Used CNN transfer learning with pre-trained AlexNet to train the dataset of 3030 images per class and achieved an accuracy of around 0.935

Implementing YOLOv5 for furniture detection (In progress)

YOLOv5 training options: python train.py --img 640 --batch 12 --epochs 30 --data dataset/fu rniture.v2-release.yolov5pytorch/data.yaml --weights yolov5m.pt --cache ram