How to Extract Detected Objects?

[Adeel-Intizar]

Hi, Can you tell me how to extract the detected objects or simply crop those objects? Thanks

[Adeel-Intizar]

I tried the following script but it gives the error, what am i missing here?

for i, bbox in enumerate(pred_bboxes):

    c_x = int(bbox[0])
    c_y = int(bbox[1])
    half_w = int(bbox[2] / 2)
    half_h = int(bbox[3] / 2)
    left = c_x - half_w
    top = c_y - half_h
    right = c_x + half_w
    bottom = c_y + half_h
    img = image.copy()[left:top, right:bottom]
    cv2.imwrite(f"crop/result-{i}.jpg", img)

[hhk7734]

pred_bboxes == Dim(-1, (x, y, w, h, class_id, probability)) and x, y, w, and h are ratios.

    height, width, _ = image.shape
    c_x = int(bbox[0] * width)
    c_y = int(bbox[1] * height)
    half_w = int(bbox[2] * width / 2)
    half_h = int(bbox[3] * height / 2)
    left = c_x - half_w
    top = c_y - half_h
    right = c_x + half_w
    bottom = c_y + half_h
    img = image.copy()[left:top, right:bottom]
    cv2.imwrite(f"crop/result-{i}.jpg", img)

[Adeel-Intizar]

I used the following script, but it still generates the error, is there something wrong with the script?

import cv2
import numpy as np
import tensorflow as tf
from yolov4.tf import YOLOv4

yolo = YOLOv4()
yolo.classes = "classes.names"
yolo.input_size = (640, 480)
yolo.batch_size = 2
dataset = yolo.load_dataset("names.txt", image_path_prefix="images/", dataset_type="yolo")

for i, (images, gt) in enumerate(dataset):
    for j in range(len(images)):
        _candidates = []
        for candidate in gt:
            grid_size = candidate.shape[1:3]
            _candidates.append(
                tf.reshape(
                    candidate[j], shape=(1, grid_size[0] * grid_size[1] * 3, -1)
                )
            )
        candidates = np.concatenate(_candidates, axis=1)

        frame = images[j, ...] * 255
        im = np.copy(frame)
        height, width, _ = im.shape
        # frame = frame.astype(np.uint8)

        bboxes = yolo.candidates_to_pred_bboxes(candidates[0])
        # bboxes = yolo.fit_pred_bboxes_to_original(pred_bboxes, frame.shape)
        if bboxes.shape[-1] == 5:
            bboxes = np.concatenate([bboxes, np.full((*bboxes.shape[:-1], 1), 2.0)], axis=-1)
        else:
            bboxes = np.copy(bboxes)

        for bbox in bboxes:

            c_x = int(bbox[0] * width)
            c_y = int(bbox[1] * height)
            half_w = int(bbox[2] * width / 2)
            half_h = int(bbox[3] * height / 2)
            left = c_x - half_w
            top = c_y - half_h
            right = c_x + half_w
            bottom = c_y + half_h
            img = im.copy()[left:top, right:bottom]
            cv2.imwrite(f"crop/result-{i}.jpg", img)

    if i == 1:
        break

cv2.destroyAllWindows()

[Adeel-Intizar]

Can you test the script on your dataset to check if it is causing error just on my dataset?

[hhk7734]

line 52 img = im[top:bottom, left:right, :]

import cv2
import numpy as np
import tensorflow as tf
from yolov4.tf import YOLOv4

yolo = YOLOv4()
yolo.classes = "/home/hhk7734/NN/coco.names"
yolo.input_size = (640, 480)
yolo.batch_size = 2

dataset = yolo.load_dataset(
    "/home/hhk7734/NN/val2017.txt",
    image_path_prefix="/home/hhk7734/NN/val2017",
    training=False,
)

for i, (images, gt) in enumerate(dataset):
    for j in range(len(images)):
        _candidates = []
        for candidate in gt:
            grid_size = candidate.shape[1:3]
            _candidates.append(
                tf.reshape(
                    candidate[j], shape=(1, grid_size[0] * grid_size[1] * 3, -1)
                )
            )
        candidates = np.concatenate(_candidates, axis=1)

        frame = images[j, ...] * 255
        im = np.copy(frame)
        height, width, _ = im.shape
        # frame = frame.astype(np.uint8)

        bboxes = yolo.candidates_to_pred_bboxes(candidates[0])
        # bboxes = yolo.fit_pred_bboxes_to_original(pred_bboxes, frame.shape)
        if bboxes.shape[-1] == 5:
            bboxes = np.concatenate(
                [bboxes, np.full((*bboxes.shape[:-1], 1), 2.0)], axis=-1
            )
        else:
            bboxes = np.copy(bboxes)

        for bbox in bboxes:
            c_x = int(bbox[0] * width)
            c_y = int(bbox[1] * height)
            half_w = int(bbox[2] * width / 2)
            half_h = int(bbox[3] * height / 2)
            left = c_x - half_w
            top = c_y - half_h
            right = c_x + half_w
            bottom = c_y + half_h
            img = im[top:bottom, left:right, :]
            cv2.imwrite(f"result-{i}.jpg", img)

    if i == 1:
        break

cv2.destroyAllWindows()

[Adeel-Intizar]

Thank you, Script is fine now :) 👍