这是一个yolox-keras的源码，可以用于训练自己的模型。

many will try using different backbones like yolo/VGG/ENet for their models and it will be usefull if you have a pretrained model that doesn't include top layers for them. I also tried to find the saved model as it has to be in this directory :'model_data/yolox_s.h5' but it isn't if you still have the model I'll appreciate to upload it again. because I tried to train the model but found some issues with utils and loading data. Thanks

!python predict.py 报错如下： model_data/yolox_s.h5 model, and classes loaded. Traceback (most recent call last): File "predict.py", line 14, in yolo = YOLO() File "/content/yolox-keras/yolo.py", line 82, in init self.boxes, self.scores, self.classes = self.generate() File "/content/yolox-keras/yolo.py", line 106, in generate letterbox_image = self.letterbox_image File "/content/yolox-keras/utils/utils_bbox.py", line 73, in DecodeBox grid_x, grid_y = tf.meshgrid(K.arange(hw[i][1]), K.arange(hw[i][0])) File "/usr/local/lib/python3.7/dist-packages/tensorflow/python/util/traceback_utils.py", line 153, in error_handler raise e.with_traceback(filtered_tb) from None File "/usr/local/lib/python3.7/dist-packages/keras/layers/core/tf_op_layer.py", line 107, in handle return TFOpLambda(op)(*args, **kwargs) File "/usr/local/lib/python3.7/dist-packages/keras/utils/traceback_utils.py", line 67, in error_handler raise e.with_traceback(filtered_tb) from None File "/usr/local/lib/python3.7/dist-packages/keras/backend.py", line 3510, in arange if stop is None and start < 0: tensorflow.python.framework.errors_impl.OperatorNotAllowedInGraphError: Exception encountered when calling layer "tf.keras.backend.arange" (type TFOpLambda).

using a tf.Tensor as a Python bool is not allowed in Graph execution. Use Eager execution or decorate this function with @tf.function.

Call arguments received: • start=tf.Tensor(shape=(), dtype=int32) • stop=None • step=1 • dtype=int32

修改：utils.utils_bbox.DecodeBox如下： @tf.function def DecodeBox()

接着报错： model_data/yolox_s.h5 model, and classes loaded. Traceback (most recent call last): File "predict.py", line 14, in yolo = YOLO() File "/content/yolox-keras/yolo.py", line 82, in init self.boxes, self.scores, self.classes = self.generate() File "/content/yolox-keras/yolo.py", line 106, in generate letterbox_image = self.letterbox_image File "/usr/local/lib/python3.7/dist-packages/tensorflow/python/util/traceback_utils.py", line 153, in error_handler raise e.with_traceback(filtered_tb) from None File "/usr/local/lib/python3.7/dist-packages/keras/engine/keras_tensor.py", line 256, in array f'You are passing {self}, an intermediate Keras symbolic input/output, ' TypeError: You are passing KerasTensor(type_spec=TensorSpec(shape=(None, None, None, 85), dtype=tf.float32, name=None), name='concatenate_13/concat:0', description="created by layer 'concatenate_13'"), an intermediate Keras symbolic input/output, to a TF API that does not allow registering custom dispatchers, such as tf.cond, tf.function, gradient tapes, or tf.map_fn. Keras Functional model construction only supports TF API calls that do support dispatching, such as tf.math.add or tf.reshape. Other APIs cannot be called directly on symbolic Kerasinputs/outputs. You can work around this limitation by putting the operation in a custom Keras layer call and calling that layer on this symbolic input/output.