How to build an Image recogniser using your own dataset
Sangho Bose
1644

Hi there, is there any way to make the pixel size bigger? 28x28 is not working for my model. When I change the pixel size, I get the following error and I’m not understanding what it means.

InvalidArgumentError (see above for traceback): Input to reshape is a tensor with 3422848 values, but the requested shape requires a multiple of 3136
 [[Node: Reshape_1 = Reshape[T=DT_FLOAT, Tshape=DT_INT32, _device=”/job:localhost/replica:0/task:0/cpu:0"](MaxPool_1, Reshape_1/shape)]]

In addition, I’m very new to tensorflow so I’m struggling to understand all of these paramets. Are the shape or padding inputs cropping each image in any way? For my model specifically, I’m trying to identify people and not people. This model works (sometimes) when the people in the model are in the center of the camera but it doesn’t work when people are on the edges of the frame.