Change size of dataset?
nattari opened this issue · 2 comments
The assumption is that there 7000 images..
How to change to the actual number of images a user has.. I have 1085 images..
I get the below error:
Check failed: num_test_image_ <= names_.size() (1497 vs. 217)
solved it by changing "num_test_image" parameter in train_ssd_kitti.py
but now i am getting this "Cannot copy param 0 weights from layer 'fc6'; shape mismatch. Source param shape is 1 1 4096 25088 (102760448); target param shape is 1024 512 3 3 (4718592). To learn this layer's parameters from scratch rather than copying from a saved net, rename the layer."
I tried changing name of field but no luck :(
@nattari a little late maybe but I had same error as you when using the "VGG16.v2.caffemodel" and when I changed to use the "VGG_ILSVRC_16_layers_fc_reduced.caffemodel" instead everything worked fine. Got the pretrained file in here:
https://github.com/conner99/VGGNet/blob/master/VGG_ILSVRC_16_layers_fc_reduced.caffemodel