Inference.py has errors in Layer #399

[cyrilzakka]

I tried running inference.py but ran into the following error:

WARNING:tensorflow:From /content/EfficientDet/layers.py:153: where (from tensorflow.python.ops.array_ops) is deprecated and will be removed in a future version.
Instructions for updating:
Use tf.where in 2.0, which has the same broadcast rule as np.where
Traceback (most recent call last):
  File "inference.py", line 64, in <module>
    main()
  File "inference.py", line 32, in main
    model.load_weights(model_path, by_name=True)
  File "/usr/local/lib/python3.6/dist-packages/tensorflow_core/python/keras/engine/training.py", line 182, in load_weights
    return super(Model, self).load_weights(filepath, by_name)
  File "/usr/local/lib/python3.6/dist-packages/tensorflow_core/python/keras/engine/network.py", line 1371, in load_weights
    saving.load_weights_from_hdf5_group_by_name(f, self.layers)
  File "/usr/local/lib/python3.6/dist-packages/tensorflow_core/python/keras/saving/hdf5_format.py", line 754, in load_weights_from_hdf5_group_by_name
    str(weight_values[i].shape) + '.')
ValueError: Layer #399 (named "class_net/class-predict"), weight <tf.Variable 'class_net/class_net/class-predict/pointwise_kernel:0' shape=(1, 1, 64, 99) dtype=float32> has shape (1, 1, 64, 99), but the saved weight has shape (1, 1, 64, 810).

Any ideas?

Here is my inference.py:
import cv2
import json
import numpy as np
import os
import time
import glob

from model import efficientdet
from utils import preprocess_image, postprocess_boxes
from utils.draw_boxes import draw_boxes

def main():
    os.environ['CUDA_VISIBLE_DEVICES'] = '0'

    phi = 0
    weighted_bifpn = False
    model_path = 'checkpoints/2020-10-18/coco_01_57.4951_82.4522.h5'
    image_sizes = (512, 640, 768, 896, 1024, 1280, 1408)
    image_size = image_sizes[phi]
    # coco classes
    classes = [
        'Class1', 'Class2', 'Class3', etc.
    ]
    num_classes = len(classes)
    score_threshold = 0.3
    colors = [np.random.randint(0, 256, 3).tolist() for _ in range(num_classes)]
    _, model = efficientdet(phi=phi,
                            weighted_bifpn=weighted_bifpn,
                            num_classes=num_classes,
                            score_threshold=score_threshold)
    model.load_weights(model_path, by_name=True)

    for image_path in glob.glob('/content/EfficientDet/datasets/coco/images/val2017/frame_00000*.PNG'):
        image = cv2.imread(image_path)
        src_image = image.copy()
        # BGR -> RGB
        image = image[:, :, ::-1]
        h, w = image.shape[:2]

        image, scale = preprocess_image(image, image_size=image_size)
        # run network
        start = time.time()
        boxes, scores, labels = model.predict_on_batch([np.expand_dims(image, axis=0)])
        boxes, scores, labels = np.squeeze(boxes), np.squeeze(scores), np.squeeze(labels)
        print(time.time() - start)
        boxes = postprocess_boxes(boxes=boxes, scale=scale, height=h, width=w)

        # select indices which have a score above the threshold
        indices = np.where(scores[:] > score_threshold)[0]

        # select those detections
        boxes = boxes[indices]
        labels = labels[indices]

        draw_boxes(src_image, boxes, scores, labels, colors, classes)

        cv2.namedWindow('image', cv2.WINDOW_NORMAL)
        cv2.imshow('image', src_image)
        cv2.waitKey(0)

if __name__ == '__main__':
    main()

I've checked, the classes match! Here is the excerpt from the JSON:

{"licenses": [{"name": "", "id": 0, "url": ""}], "info": {"contributor": "", "date_created": "", "description": "", "url": "", "version": "", "year": ""}, "categories": [{"id": 1, "name": "Class1", "supercategory": ""}, {"id": 2, "name": "Class2", "supercategory": ""}, {"id": 3, "name": "Class3", "supercategory": ""}, {"id": 4, "name": "Class4", "supercategory": ""}, {"id": 5, "name": "Class5", "supercategory": ""}, {"id": 6, "name": "Class6", "supercategory": ""}, {"id": 7, "name": "Class7", "supercategory": ""}, {"id": 8, "name": "Class8", "supercategory": ""}, {"id": 9, "name": "Class9", "supercategory": ""}, {"id": 10, "name": "Class10", "supercategory": ""}, {"id": 11, "name": "Class11", "supercategory": ""}]

[cyrilzakka]

I've just realized this might be because when training, the number of classes was never specified. Is there a way of specifying the number of classes for the training script?

[cyrilzakka]

Alright so this officially worked! Unfortunately runninginference.py still returns the image with no bounding boxes at all.

[datleeee]

Alright so this officially worked! Unfortunately runninginference.py still returns the image with no bounding boxes at all.

Hi, I'm having the same problem. How did you solve this? Thank you.