the Output image size?

[tuqiangwang]

Original image and heatmap are inconsistent in size
why???

[iamaaditya]

Heatmap size is dictated by the network. As this network is not fully-convolutional it outputs a fixed size heatmaps. Heatmap can be resized to original image. Thanks and Regards Adi

…

On Thu, Apr 4, 2019 at 4:56 AM ByYourSide ***@***.***> wrote: Original image and heatmap are inconsistent in size why??? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#13>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AB008rvWkYN4xB05aNYDBBW3g9JkOGXEks5vdb46gaJpZM4ccRQT> .

[tuqiangwang]

How to adjust it? Do you adjust during training or during testing

[iamaaditya]

You do not have to adjust it. Just resize the heatmap to the size of the image. Thanks and Regards Adi

…

On Fri, Apr 5, 2019 at 11:46 AM ByYourSide ***@***.***> wrote: How to adjust it? Do you adjust during training or during testing — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#13 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AB008oEJ2ouBdy640yGI4IdlI47OSvZkks5vd2_agaJpZM4ccRQT> .

[tuqiangwang]

But after this adjustment, the content of the heat map and the content of the original map is not a corresponding one. In other words, there is no corresponding heat map output in the original picture

[iamaaditya]

Then probably your image is too large and with different aspect ratio. Can you crop your image to the correct aspect ratio, so that after resizing all the objects are in the same region? Thanks and Regards Adi

…

On Fri, Apr 5, 2019 at 11:51 AM ByYourSide ***@***.***> wrote: But after this adjustment, the content of the heat map and the content of the original map is not a corresponding one. In other words, there is no corresponding heat map output in the original picture — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#13 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AB008s92Ugc18PEMoHZxdf1HLN9msCwAks5vd3DygaJpZM4ccRQT> .

[tuqiangwang]

Thanks
What is the aspect ratio that your project supports? As an aside, I don't think deep learning does a good job of compressing images. It's not nearly as good as Jpeg, like Google. What do you think?

[iamaaditya]

I am not sure but I think it supports square image (1:1). What do you mean it is not better than JPEG. This project uses JPEG to compress images. So it should be at least as good as standard JPEG. Thanks and Regards Adi

…

On Fri, Apr 5, 2019 at 11:57 AM ByYourSide ***@***.***> wrote: Thanks What is the aspect ratio that your project supports? As an aside, I don't think deep learning does a good job of compressing images. It's not nearly as good as Jpeg, like Google. What do you think? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#13 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AB008pWw7bQcPd6Da5uoE1B_WwXAr_klks5vd3J9gaJpZM4ccRQT> .

[tuqiangwang]

I just can't guarantee that my input image will produce a heat map of the same size, can I say that? And, as far as you know, there are other GitHub projects that generate heat maps