Created by: felixgalaxy

用网络爬取的数据集单张图80k左右，训练都正常 但是用自己拍的数据集训练，单张图片2m，一共300多张图，训练报错，是因为图片太大吗？请问应该怎么处理呢，如果裁剪或者resize了我标注的xml文件是不是不可以使用了？求助 2020-06-01 18:38:51,718-WARNING: recv endsignal from outq with errmsg[consumer[consumer-bab-2] failed to map with error:[could not broadcast input array from shape (4032,3024,3) into shape (3024,4032,3)]] 2020-06-01 18:39:04,304-INFO: fail to map op [RandomExpand_d635b1] with error: could not broadcast input array from shape (4032,3024,3) into shape (3024,4032,3) and stack: Traceback (most recent call last): File "/public/home/felix/myproject/Paddle/PaddleDetection/ppdet/data/reader.py", line 45, in call data = f(data, ctx) File "/public/home/felix/myproject/Paddle/PaddleDetection/ppdet/data/transform/operators.py", line 1268, in call canvas[y:y + height, x:x + width, :] = img.astype(np.uint8) ValueError: could not broadcast input array from shape (4032,3024,3) into shape (3024,4032,3)

Error Message Summary:

Error: Blocking queue is killed because the data reader raises an exception [Hint: Expected killed_ != true, but received killed_:1 == true:1.] at (/paddle/paddle/fluid/operators/reader/blocking_queue.h:141) [operator < read > error]