Object Detection: YOLO & Faster-RCNN

# pip install ultralytics
from ultralytics import YOLO
import numpy as np
import cv2

# ── 1. Load pretrained YOLO v8 ────────────────────────────────────────────────
model = YOLO("yolov8n.pt")   # nano model (3.2M params, fastest)
# Other sizes: yolov8s.pt, yolov8m.pt, yolov8l.pt, yolov8x.pt

# ── 2. Inference on a single image ────────────────────────────────────────────
results = model("path/to/image.jpg", conf=0.25, iou=0.5)

for r in results:
    boxes = r.boxes                  # Boxes object
    for box in boxes:
        x1, y1, x2, y2 = box.xyxy[0].tolist()  # absolute pixel coords
        conf  = box.conf[0].item()              # confidence score
        cls   = int(box.cls[0].item())          # class index
        label = model.names[cls]
        print(f"{label}: {conf:.2f} at ({x1:.0f},{y1:.0f},{x2:.0f},{y2:.0f})")

# ── 3. Fine-tuning on custom dataset ─────────────────────────────────────────
# Dataset format: YOLO txt format
# data.yaml:
#   train: /path/to/train/images
#   val:   /path/to/val/images
#   nc: 3                       # number of classes
#   names: ['cat', 'dog', 'car']

model = YOLO("yolov8s.pt")     # start from ImageNet pretrained
results = model.train(
    data="data.yaml",
    epochs=50,
    imgsz=640,
    batch=16,
    lr0=0.01,                   # initial learning rate
    lrf=0.01,                   # final lr fraction
    augment=True,               # mosaic, flip, scale augmentation
    device=0,                   # GPU 0
)
print(f"mAP50: {results.metrics.mAP50:.4f}")

# ── 4. IoU calculation from scratch ──────────────────────────────────────────
def iou(box1, box2):
    """box = [x1, y1, x2, y2]"""
    x1 = max(box1[0], box2[0]); y1 = max(box1[1], box2[1])
    x2 = min(box1[2], box2[2]); y2 = min(box1[3], box2[3])
    inter = max(0, x2-x1) * max(0, y2-y1)
    area1 = (box1[2]-box1[0]) * (box1[3]-box1[1])
    area2 = (box2[2]-box2[0]) * (box2[3]-box2[1])
    return inter / (area1 + area2 - inter + 1e-6)

gt   = [100, 50, 250, 200]
pred = [110, 60, 260, 210]
print(f"\nIoU = {iou(gt, pred):.4f}")

# ── 5. Manual NMS ─────────────────────────────────────────────────────────────
def nms(boxes, scores, iou_threshold=0.5):
    """Boxes: (N,4) xyxy, Scores: (N,)"""
    order = np.argsort(scores)[::-1]
    keep  = []
    while len(order) > 0:
        i = order[0]
        keep.append(i)
        ious = np.array([iou(boxes[i], boxes[j]) for j in order[1:]])
        order = order[1:][ious < iou_threshold]
    return keep

boxes  = np.array([[100,50,250,200],[105,55,255,205],[200,100,350,250]])
scores = np.array([0.95, 0.87, 0.72])
kept   = nms(boxes, scores)
print(f"Kept boxes: {kept}")  # [0, 2] — box 1 suppressed (overlaps with 0)