CocoPi/yolo2_20class_awnn.py

from maix import nn, camera, image, display
from maix.nn import decoder
import time

model = {
    "param": "/home/model/yolo2_20class_awnn.param",
    "bin": "/home/model/yolo2_20class_awnn.bin"
}
options = {
    "model_type":  "awnn",
    "inputs": {
        "input0": (224, 224, 3)
    },
    "outputs": {
        "output0": (7, 7, (1+4+20)*5)
    },
    "mean": [127.5, 127.5, 127.5],
    "norm": [0.0078125, 0.0078125, 0.0078125],
}

labels = ["aeroplane", "bicycle", "bird", "boat", "bottle", "bus", "car", "cat", "chair", "cow", "diningtable", "dog", "horse", "motorbike", "person", "pottedplant", "sheep", "sofa", "train", "tvmonitor"]
anchors = [5.4, 5.38, 1.65, 2.09, 0.8, 1.83, 2.45, 4.14, 0.46, 0.8]

m = nn.load(model, opt=options)
yolo2_decoder = decoder.Yolo2(len(labels), anchors, net_in_size=(options["inputs"]["input0"][0], options["inputs"]["input0"][1]), net_out_size=(7, 7))

while True:
    img = camera.capture()
    AI_img = img.copy().resize(224, 224)
    out = m.forward(AI_img.tobytes(), quantize=True, layout="hwc")
    boxes, probs = yolo2_decoder.run(out, nms=0.3, threshold=0.3, img_size=(options["inputs"]["input0"][0], options["inputs"]["input0"][1]))

    if len(boxes):
        for i, box in enumerate(boxes):
            class_id = probs[i][0]
            prob = probs[i][1][class_id]
            disp_str = "{}:{:.2f}%".format(labels[class_id], prob*100)
            img.draw_rectangle(box[0], box[1], box[0] + box[2], box[1] + box[3], color = (0, 255, 0), thickness=2)
            x = box[0]
            y = box[1] - 20
            if y < 0:
                y = 0
            img.draw_string(x, y, disp_str, 1.5, color = (255, 0, 0), thickness=2)

    display.show(img)