PaddleSeg-2.6.0中运行官方演示代码出错解决办法

---

问题描述

Q1：

在运行时，出现了以下的警告：I0706 13:09:12.768538 3772 eager_method.cc:140] Warning:: 0D Tensor cannot be used as 'Tensor.numpy()[0]' . In order to avoid this problem, 0D Tensor will be changed to 1D numpy currently, but it's not correct and will be removed in release 2.6. For Tensor contain only one element, Please modify 'Tensor.numpy()[0]' to 'float(Tensor)' as soon as possible, otherwise 'Tensor.numpy()[0]' will raise error in release 2.6.

A1：

训练前先使用‘export FLAGS_set_to_1d=False’命令，意为忽略警告。

---

Q2：

在模型的验证时为什么会报错误：

2023-07-12 14:26:44 [INFO] [TRAIN] epoch: 17, iter: 500/10000, loss: 1.8821, lr: 0.000972, batch_cost: 0.4265, reader_cost: 0.30989, ips: 14.0681 samples/sec | ETA 01:07:31
2023-07-12 14:26:44 [INFO] Start evaluating (total_samples: 45, total_iters: 45)...
Traceback (most recent call last):
File "/home/aistudio/PaddleSeg/tools/train.py", line 195, in
main(args)
File "/home/aistudio/PaddleSeg/tools/train.py", line 170, in main
train(
File "/opt/conda/envs/python35-paddle120-env/lib/python3.10/site-packages/paddleseg/core/train.py", line 315, in train
mean_iou, acc, _, _, _ = evaluate(
File "/opt/conda/envs/python35-paddle120-env/lib/python3.10/site-packages/paddleseg/core/val.py", line 161, in evaluate
intersect_area, pred_area, label_area = metrics.calculate_area(
File "/opt/conda/envs/python35-paddle120-env/lib/python3.10/site-packages/paddleseg/utils/metrics.py", line 57, in calculate_area
pred_area = paddle.concat(pred_area)
File "/opt/conda/envs/python35-paddle120-env/lib/python3.10/site-packages/paddle/tensor/manipulation.py", line 1121, in concat
return _C_ops.concat(input, axis)
ValueError: (InvalidArgument) The axis is expected to be in range of [0, 0), but got 0
[Hint: Expected axis >= -rank && axis < rank == true, but received axis >= -rank && axis < rank:0 != true:1.] (at ../paddle/phi/infermeta/multiary.cc:961)

A2：

1./home/aistudio/PaddleSeg/paddleseg/core/train.py 文件内'avg_loss += loss.numpy()[0]'改成‘avg_loss += loss.numpy()’
2.‘avg_loss_list = [l[0] / log_iters for l in avg_loss_list]’也是要改成‘avg_loss_list = [l / log_iters for l in avg_loss_list]’。
3.metrics.py文件里的‘pred_area.append(paddle.sum(paddle.cast(pred_i, "int32")))’也要改成‘pred_area.append(paddle.sum(paddle.cast(pred_i, "int32")).unsqueeze(0))’
这样才能正常运行。

---

Q3：

使用VOCaug数据集时会出现错误。

A3：

在配置文件中去掉aug，不使用增强VOC数据集可以正常运行。

将/home/aistudio/PaddleSeg/configs/deeplabv3p/deeplabv3p_resnet50_os8_voc12aug_512x512_40k.yml文件里的‘base: '../base/pascal_voc12aug.yml'修改为了‘base: '../base/pascal_voc12.yml'’。

---