maju116/yolo3-coco-keras

YOLOv3 COCO – Keras 3

YOLOv3 object detection model trained on COCO dataset.

Framework

• Keras 3
• TensorFlow backend

Input

• Shape: (416, 416, 3)
• RGB images, values in [0, 1] or [0, 255]

Output

Model returns three tensors:

• grid1: (13, 13, 3, 85)
• grid2: (26, 26, 3, 85)
• grid3: (52, 52, 3, 85)

(85 = 4 bbox + 1 objectness + 80 classes)

Load example

from keras.models import load_model

model = load_model("coco_yolo_v3.keras", compile=False)