user@devops:~$ cat README.md
Computer Vision Advanced
# Description
Complete computer vision project with two fundamental models. YOLOv8n (Ultralytics) for object detection: single-shot CNN architecture, 3.2M parameters, 80 COCO classes, bounding boxes with confidence, runs on CPU across 3 sample images (urban street, nature landscape, indoor room). DeepLabV3-MobileNetV3 (Torchvision) for semantic segmentation: Atrous Convolution + ASPP module, 11M parameters, 21 Pascal VOC classes, pixel-wise map with colored overlay. 3x3 comparative panel showing Original / YOLO / Segmentation for each image. Includes detailed technical specification with key differences between detection and segmentation.
# Key features
$ YOLOv8n: single-shot detection with bounding boxes and confidence (80 COCO classes)
$ DeepLabV3: pixel-wise semantic segmentation (21 VOC classes)
$ Pre-trained models running on CPU without GPU
$ 3 sample images: urban, nature, indoor
$ 3x3 comparison panel: Original vs YOLO vs Segmentation
$ Individual visualizations saved to docs/
$ Technical spec with architecture, metrics, and key differences
# Gallery
# Technologies used
