`TypeError: unsupported operand type(s) for *. 'Operation' and 'int': 'Operation'和'int'`。Keras tensorflow

Question

我正试图根据这里的代码实现yolov3模型(https:/github.comexperiencorkeras-yolo3blobmasteryolo.py。)

这段代码是用tensor-flow 1.x写的，我想把它转换成2.0。我正在尝试做所有需要的改变，例如 tf.Print to tf.print 等，但卡在了192行 return loss*self.grid_scale. 我得到的错误是 TypeError: unsupported operand type(s) for *: 'Operation' and 'int'.我如何在 loss. 目前，我得到的类型操作回。我试图获取操作中的值，我已经尝试了 execute_eagerly,.eval 但仍然无法获得操作值。

完整的代码在上面的链接中。而主要的片段是

 loss = tf.print('avg_obj \t\t:',loss, [grid_h, avg_obj],output_stream=sys.stderr)
        loss = tf.print(loss, [grid_h, avg_noobj], 'avg_noobj \t\t', summarize=1000)
        loss = tf.print(loss, [grid_h, avg_iou],  'avg_iou \t\t', summarize=1000)
        loss = tf.print(loss, [grid_h, avg_cat],  'avg_cat \t\t', summarize=1000)
        loss = tf.print(loss, [grid_h, recall50],  'recall50 \t', summarize=1000)
        loss = tf.print(loss, [grid_h, recall75], 'recall75 \t', summarize=1000)   
        loss = tf.print(loss, [grid_h, count],  'count \t', summarize=1000)     
        loss = tf.print(loss, [grid_h, tf.reduce_sum(loss_xy), 
                                       tf.reduce_sum(loss_wh), 
                                       tf.reduce_sum(loss_conf), 
                                       tf.reduce_sum(loss_class)],   'loss xy, wh, conf, class: \t',   summarize=1000) 

        return loss*self.grid_scale

以下是完整的错误跟踪

回溯（最近一次调用）。

文件""，第1行，在runfile('C:UsersSreeDocumentskeras-yolo3-mastertrain.py', wdir='C:UsersSreeDocumentskeras-yolo3-master')

文件 "C:\Users\Sree\Anaconda3\envs\tensorflow2\lib\site-packages\spyder_kernels\customize\spydercustomize.py"，第827行，在运行文件execfile(filename, namespace) 中。

文件 "C:\Users\Sree\Anaconda3\envs\tensorflow2\lib\site-packages\spyder_kernels\customize\spydercustomize.py"，第110行，在execfile exec(compile(f.read(), filename, 'exec'), namespace中。)

文件 "C:UsersSreeDocumentskeras-yolo3-mastertrain.py"，第279行，在 主(args)

文件 "C:UsersSreeDocumentskeras-yolo3-mastertrain.py"，第241行，在 主 class_scale = config['train']['class_scale'],

文件 "C:UsersSreeDocumentskeras-yolo3-mastertrain.py"，第131行，在create_model中，class_scale = class_scale。

文件 "C:\Users\SreeDocuments\keras-yolo3-master\yolo.py", 第320行, in create_yolov3_model class_scale)([input_image, pred_yolo_1, true_yolo_1, true_boxes])

文件 "C:\Users\Sree\Anaconda3\envs\tensorflow2\lib\sit-packages\keras/backend\tensorflow_backend.py", 第75行, 在 symbolic_fn_wrapper return func(*args, **kwargs)中。

文件 "C:/Users/SreeAnaconda3/envs/tensorflow2/lib/site-packages/kerasengine/base_layer.py"，第489行，in 召唤 output = self.call(inputs, **kwargs)

文件 "C:\Users\SreeDocuments\keras-yolo3-master\yolo.py"，第206行，在调用返回损失*self.grid_scale中

TypeError: unsupported operand type(s) for *: 'Operation'和'int'

Answer 1

你需要像下面提到的Tensorflow 2.0中的代码一样定义损失。

def YoloLoss(anchors, classes=80, ignore_thresh=0.5):
    def yolo_loss(y_true, y_pred):
        # 1. transform all pred outputs
        # y_pred: (batch_size, grid, grid, anchors, (x, y, w, h, obj, ...cls))
        pred_box, pred_obj, pred_class, pred_xywh = yolo_boxes(
            y_pred, anchors, classes)
        pred_xy = pred_xywh[..., 0:2]
        pred_wh = pred_xywh[..., 2:4]

        # 2. transform all true outputs
        # y_true: (batch_size, grid, grid, anchors, (x1, y1, x2, y2, obj, cls))
        true_box, true_obj, true_class_idx = tf.split(
            y_true, (4, 1, 1), axis=-1)
        true_xy = (true_box[..., 0:2] + true_box[..., 2:4]) / 2
        true_wh = true_box[..., 2:4] - true_box[..., 0:2]

        # give higher weights to small boxes
        box_loss_scale = 2 - true_wh[..., 0] * true_wh[..., 1]

        # 3. inverting the pred box equations
        grid_size = tf.shape(y_true)[1]
        grid = tf.meshgrid(tf.range(grid_size), tf.range(grid_size))
        grid = tf.expand_dims(tf.stack(grid, axis=-1), axis=2)
        true_xy = true_xy * tf.cast(grid_size, tf.float32) - \
            tf.cast(grid, tf.float32)
        true_wh = tf.math.log(true_wh / anchors)
        true_wh = tf.where(tf.math.is_inf(true_wh),
                           tf.zeros_like(true_wh), true_wh)

        # 4. calculate all masks
        obj_mask = tf.squeeze(true_obj, -1)
        # ignore false positive when iou is over threshold
        best_iou = tf.map_fn(
            lambda x: tf.reduce_max(broadcast_iou(x[0], tf.boolean_mask(
                x[1], tf.cast(x[2], tf.bool))), axis=-1),
            (pred_box, true_box, obj_mask),
            tf.float32)
        ignore_mask = tf.cast(best_iou < ignore_thresh, tf.float32)

        # 5. calculate all losses
        xy_loss = obj_mask * box_loss_scale * \
            tf.reduce_sum(tf.square(true_xy - pred_xy), axis=-1)
        wh_loss = obj_mask * box_loss_scale * \
            tf.reduce_sum(tf.square(true_wh - pred_wh), axis=-1)
        obj_loss = binary_crossentropy(true_obj, pred_obj)
        obj_loss = obj_mask * obj_loss + \
            (1 - obj_mask) * ignore_mask * obj_loss
        # TODO: use binary_crossentropy instead
        class_loss = obj_mask * sparse_categorical_crossentropy(
            true_class_idx, pred_class)

        # 6. sum over (batch, gridx, gridy, anchors) => (batch, 1)
        xy_loss = tf.reduce_sum(xy_loss, axis=(1, 2, 3))
        wh_loss = tf.reduce_sum(wh_loss, axis=(1, 2, 3))
        obj_loss = tf.reduce_sum(obj_loss, axis=(1, 2, 3))
        class_loss = tf.reduce_sum(class_loss, axis=(1, 2, 3))

        return xy_loss + wh_loss + obj_loss + class_loss
    return yolo_loss 

对于Tensorflow 2.0中的详细代码，你可以按照这个来做。衔接。