adityac94 / Grad_CAM_plus_plus

A generalized gradient-based CNN visualization technique
284 stars 56 forks source link

show heatmaps for 3DCNN #7

Open emily19941102 opened 5 years ago

emily19941102 commented 5 years ago

I ran through your code and found that the map is dark. How can it become a heat map?thanks! @adityac94

adityac94 commented 5 years ago

Hi, Which code did you exactly run, for 3DCNNs? You mean the code on GitHub gives dark heat maps?

Thank you, Aditya

On Apr 24, 2019, at 23:34, emily19941102 notifications@github.com wrote:

I ran through your code and found that the map is dark. How can it become a heat map?thanks! @adityac94 https://github.com/adityac94 — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/adityac94/Grad_CAM_plus_plus/issues/7, or mute the thread https://github.com/notifications/unsubscribe-auth/AG3XGILVYLPTCNWGX6WGXRDPSERFFANCNFSM4HIJPEGA.

emily19941102 commented 5 years ago

Sorry, I just found out that the code is test_model.py. The prominent part of the map is white and the rest is black.I want to turn it into a colorful heat map, thanks! @adityac94

anirbansarkar-cs commented 5 years ago

@emily19941102 ...while saving the frames, to turn it into colorful heat map, you can replace the line vid_gd_img_mod = X[i] \times vid_cam_mod with the below line vid_gd_img_mod = cv2.applyColorMap(np.uint8(255 \times vid_cam_mod), cv2.COLORMAP_JET)

emily19941102 commented 5 years ago

@tataiani wow ,Thank you very much!

leelbc commented 3 years ago

@emily19941102 ...while saving the frames, to turn it into colorful heat map, you can replace the line vid_gd_img_mod = X[i] \times vid_cam_mod with the below line vid_gd_img_mod = cv2.applyColorMap(np.uint8(255 \times vid_cam_mod), cv2.COLORMAP_JET)

Hi, I'm sorry to bother you, I found that the map is still black and white after I modified it according to your code

leelbc commented 3 years ago

@emily19941102 ...while saving the frames, to turn it into colorful heat map, you can replace the line vid_gd_img_mod = X[i] \times vid_cam_mod with the below line vid_gd_img_mod = cv2.applyColorMap(np.uint8(255 \times vid_cam_mod), cv2.COLORMAP_JET)

Hi, I'm sorry to bother you, I found that the map is still black and white after I modified it according to your code

Sorry, it was due to my carelessness, the problem has been solved now, thanks

JackDanHollister commented 1 year ago

Did you get the heat map working with the 3DCNN in the end @leelbc? I am interested as I am about to try it

emily19941102 commented 1 year ago

    您好,我已收到您的邮件。我会尽快处理,给您回复。