Syed Muhmmad Usman Ghani comments

Results 18 comments of


                                            Syed Muhmmad Usman Ghani

Negative dimension size caused by subtracting 3 from 2 for '{{node max_pooling2d/MaxPool}} = MaxPool[T=DT_FLOAT, data_format="NHWC", ksize=[1, 3, 3, 1], padding="VALID", strides=[1, 2, 2, 1]](activation/Relu)' with input shapes: [?,2,112,64].

No it did not occur while i was using it Sent from Outlook ________________________________ From: Alex Mihale Sent: Sunday, August 16, 2020 4:18 PM To: foamliu/Car-Recognition Cc: Subscribed Subject: Re:...

Negative dimension size caused by subtracting 3 from 2 for '{{node max_pooling2d/MaxPool}} = MaxPool[T=DT_FLOAT, data_format="NHWC", ksize=[1, 3, 3, 1], padding="VALID", strides=[1, 2, 2, 1]](activation/Relu)' with input shapes: [?,2,112,64].

Python: 3.6.5 Tensorflow : 1.13.1 Tensorflow-gpu(if using gpu) : 1.13.1 Keras: 2.0.0 Sent from Outlook ________________________________ From: Alex Mihale Sent: Monday, August 17, 2020 2:16 PM To: foamliu/Car-Recognition Cc: Usman-Ghani123...

Negative dimension size caused by subtracting 3 from 2 for '{{node max_pooling2d/MaxPool}} = MaxPool[T=DT_FLOAT, data_format="NHWC", ksize=[1, 3, 3, 1], padding="VALID", strides=[1, 2, 2, 1]](activation/Relu)' with input shapes: [?,2,112,64].

Not at all. It worked completely fine you can try for numpy ==1.16.1 version as well with all these requirement i send you formerly Sent from Outlook ________________________________ From: Alex...

Negative dimension size caused by subtracting 3 from 2 for '{{node max_pooling2d/MaxPool}} = MaxPool[T=DT_FLOAT, data_format="NHWC", ksize=[1, 3, 3, 1], padding="VALID", strides=[1, 2, 2, 1]](activation/Relu)' with input shapes: [?,2,112,64].

No it can not work on tensorflow greater then 2. Get Outlook for Android ________________________________ From: Alex Mihale Sent: Thursday, August 27, 2020 12:17:27 PM To: foamliu/Car-Recognition Cc: Syed Muhmmad...

Negative dimension size caused by subtracting 3 from 2 for '{{node max_pooling2d/MaxPool}} = MaxPool[T=DT_FLOAT, data_format="NHWC", ksize=[1, 3, 3, 1], padding="VALID", strides=[1, 2, 2, 1]](activation/Relu)' with input shapes: [?,2,112,64].

It can only work on tensorflow 1.13.1 Get Outlook for Android ________________________________ From: Syed Muhammad Usman Ghani Sent: Thursday, August 27, 2020 1:11:29 PM To: foamliu/Car-Recognition Subject: Re: [foamliu/Car-Recognition] Negative...

Negative dimension size caused by subtracting 3 from 2 for '{{node max_pooling2d/MaxPool}} = MaxPool[T=DT_FLOAT, data_format="NHWC", ksize=[1, 3, 3, 1], padding="VALID", strides=[1, 2, 2, 1]](activation/Relu)' with input shapes: [?,2,112,64].

No, I do not know any car recognition project with TensorFlow 2+

Negative dimension size caused by subtracting 3 from 2 for '{{node max_pooling2d/MaxPool}} = MaxPool[T=DT_FLOAT, data_format="NHWC", ksize=[1, 3, 3, 1], padding="VALID", strides=[1, 2, 2, 1]](activation/Relu)' with input shapes: [?,2,112,64].

Which code are you trying to run?

Negative dimension size caused by subtracting 3 from 2 for '{{node max_pooling2d/MaxPool}} = MaxPool[T=DT_FLOAT, data_format="NHWC", ksize=[1, 3, 3, 1], padding="VALID", strides=[1, 2, 2, 1]](activation/Relu)' with input shapes: [?,2,112,64].

If you want to use TensorFlow 2+ so then you have to do some code changes in the files according to the errors

Negative dimension size caused by subtracting 3 from 2 for '{{node max_pooling2d/MaxPool}} = MaxPool[T=DT_FLOAT, data_format="NHWC", ksize=[1, 3, 3, 1], padding="VALID", strides=[1, 2, 2, 1]](activation/Relu)' with input shapes: [?,2,112,64].

In pull request, a person has suggested some changes to make this compatible with Tensorflow 2.x .

Convert into tflite

Have you downloaded the respective weight file? bcoz the weight file link is given in readme section