def main(_argv): if FLAGS.tiny: yolo = YoloV3Tiny(classes=FLAGS.num_classes) else: yolo = YoloV3(classes=FLAGS.num_classes) yolo.summary() logging.info('model created') load_darknet_weights(yolo, FLAGS.weights, FLAGS.tiny) logging.info('weights loaded') img = np.random.random((1, 320, 320, 3)).astype(np.float32) output = yolo(img) logging.info('sanity check passed') yolo.save_weights(FLAGS.output) logging.info('weights saved')
def main(_argv): physical_devices = tf.config.experimental.list_physical_devices('GPU') if len(physical_devices) > 0: tf.config.experimental.set_memory_growth(physical_devices[0], True) yolo = yolo_v3(classes=FLAGS.num_classes) yolo.summary() logging.info('model created') load_darknet_weights(yolo, FLAGS.weights, FLAGS.tiny) logging.info('weights loaded') img = np.random.random((1, 320, 320, 3)).astype(np.float32) output = yolo(img) logging.info('sanity check passed') yolo.save_weights(FLAGS.output) logging.info('weights saved')
def main(_argv): if tf.test.is_gpu_available(): tf.config.gpu.set_per_process_memory_fraction(FLAGS.gpu_fraction) if FLAGS.tiny: yolo = YoloV3Tiny() else: yolo = YoloV3() yolo.summary() logging.info('model created') load_darknet_weights(yolo, FLAGS.weights, FLAGS.tiny) logging.info('weights loaded') img = np.random.random((1, 320, 320, 3)).astype(np.float32) output = yolo(img) logging.info('sanity check passed') yolo.save_weights(FLAGS.output) logging.info('weights saved')
def main(_argv): config = tf.compat.v1.ConfigProto() config.gpu_options.allow_growth = True tf.compat.v1.InteractiveSession(config=config) if FLAGS.tiny: yolo = YoloV3Tiny(classes=FLAGS.num_classes) else: yolo = YoloV3(classes=FLAGS.num_classes) yolo.summary() logging.info('model created') load_darknet_weights(yolo, FLAGS.weights, FLAGS.tiny) logging.info('weights loaded') img = np.random.random((1, 320, 320, 3)).astype(np.float32) output = yolo(img) logging.info('sanity check passed') yolo.save_weights(FLAGS.output) logging.info('weights saved')
def main(_argv): if tf_version == '2': physical_devices = tf.config.experimental.list_physical_devices('GPU') if len(physical_devices) > 0: tf.config.experimental.set_memory_growth(physical_devices[0], True) size = None if FLAGS.size == 0 else FLAGS.size if FLAGS.tiny: yolo = YoloV3Tiny(size=size, classes=FLAGS.num_classes, training=FLAGS.training) elif FLAGS.lite: yolo = YoloV2Lite(size=size, classes=FLAGS.num_classes, training=FLAGS.training) else: yolo = YoloV3(size=size, classes=FLAGS.num_classes, training=FLAGS.training) yolo.summary() logging.info('model created') if FLAGS.from_h5: yolo.load_weights(FLAGS.weights) else: load_darknet_weights(yolo, FLAGS.weights, FLAGS.tiny, FLAGS.lite) logging.info('weights loaded') imsize = 320 if size is None else size img = np.random.random((1, imsize, imsize, 3)).astype(np.float32) output = yolo(img) logging.info('sanity check passed') predict(yolo) if FLAGS.output.endswith(('.h5', '.hdf5')): yolo.save(FLAGS.output) logging.info('keras model saved') else: yolo.save_weights(FLAGS.output) logging.info('weights saved') if tf_version == '1': convert_to_pb(yolo) logging.info('pb saved')
def convert_weights_to_tf(): if FLAGS.tiny: yolo = YoloV3Tiny(classes=80) load_darknet_weights(model=yolo, weights_file=FLAGS.tiny_weights, tiny=FLAGS.tiny) logging.info('weights loaded') img = np.random.random((1, 320, 320, 3)).astype(np.float32) output = yolo(img) logging.info('sanity check passed') yolo.save_weights('./checkpoints/yolov3-tiny.tf') logging.info('weights saved') else: yolo = YoloV3(classes=80) load_darknet_weights(model=yolo, weights_file=FLAGS.weights, tiny=FLAGS.tiny) logging.info('weights loaded') img = np.random.random((1, 320, 320, 3)).astype(np.float32) output = yolo(img) logging.info('sanity check passed') yolo.save_weights(FLAGS.output) logging.info('weights saved')
import numpy as np from yolov3_tf2.models_simple import YoloV3, YoloV3Tiny from yolov3_tf2.utils import load_darknet_weights import tensorflow as tf tiny = False num_classes = 80 path_weights = './data/yolov3.weights' path_output = './checkpoints/yolov3.tf' if tiny: path_weights = path_weights[:-8] + '-tiny.weights' path_output = path_output[:-3] + '-tiny.tf' physical_devices = tf.config.experimental.list_physical_devices('GPU') if len(physical_devices) > 0: tf.config.experimental.set_memory_growth(physical_devices[0], True) if tiny: yolo = YoloV3Tiny(classes=num_classes) else: yolo = YoloV3(classes=num_classes) yolo.summary() load_darknet_weights(yolo, path_weights, tiny) img = np.random.random((1, 320, 320, 3)).astype(np.float32) output = yolo(img) yolo.save_weights(path_output)