Hello, there was an error when I tried to run the training classifier.

[Hong409]

UnknownError: Failed to get convolution algorithm. This is probably because cuDNN failed to initialize, so try looking to see if a warning log message was printed above. [[node vgg16/block1_conv1/Conv2D (defined at D:\Anaconda+opencv\lib\site-packages\tensorflow_core\python\framework\ops.py:1751) ]] [Op:__inference_distributed_function_1536]

I used tensorflow2.0.0 cuda10.0 cudnn 7.4.0. Can you tell me your configuration environment if possible

[madeyoga]

Hai Hong409.

Here is the exact version of tf I used for training & testing the classifier.

tensorboard==2.3.0
tensorboard-plugin-wit==1.7.0
tensorflow-gpu==2.3.0
tensorflow-gpu-estimator==2.3.0

• cuda 10.1
• cudnn 7.6.5

Note

The classifier actually does not have a good performance (failed) when tested for object detection (detecting bubbles).

So I changed the dataset, using Manga109 Dataset and use darknet with YoloV3 configuration to detect the dialogue texts instead.

You can get the weights and the configuration here, and use the code example to detect the dialogue texts. https://github.com/madeyoga/detect-bubble#darknet-yolov3-object-detection--manga109-dataset