nan value in evaluate_mAP.py

[hukutonpl]

t1 = time.time()
if YOLO_FRAMEWORK == "tf":
    pred_bbox = Yolo.predict(image_data)

predict gives nan value after the first image.
This gives very low mAP

[hukutonpl]

Change it to this

t1 = time.time()
if YOLO_FRAMEWORK == "tf":
        Yolo.load_weights(f"./checkpoints/{TRAIN_MODEL_NAME}")
        pred_bbox = Yolo.predict(image_data)