baristahell

Well, i'm definitely interested, have you managed to build a working image?

Seems like it mixes up the prototxts and caffe-grid-0.1-SNAPSHOT-jar-with-dependencies.jar What are your ENV variables? What kind of cluster are you working on? The default from the wiki page? EDIT :...

I feel like CaffeOnSpark on cluster other than Yarn ones is really rigid. You have to make sure that `clustersize*executorcores=coresmax`, while `batchsize>=trainsteps*devices*clustersize`, at least this works for me

@ahsan856jalal is, like often, refering to @AlexeyAB 's fork : [AlexeyAB/darknet](https://github.com/AlexeyAB/darknet)

You have to get the data from [there](http://www.cvlibs.net/datasets/kitti/). The paths in the trainval prototxt will be the path to `[...]/kitti_data/window_files/mscnn_window_file_kitti_vehicle_train.txt`, and this txt file contains the path to every single...

It's simply the [`left color images of object data set`] part, but you can also get [`training labels of object data set`], in order to have clean {image;labels} couples. You'll...

I think, if i recall correctly, that i had this kind of issues and had to switch to cudnn v3 But not sure, it makes a while, wait for someone...

Is it related to this [comment](https://github.com/xingwangsfu/caffe-yolo/issues/7#issuecomment-249029623) ? I just can't convert my yolo.weights, same issue than number 7, and it might be related..

"num" is the number of anchors you're asking for. It's divided by 3 in YOLOv3 because the network generates 3 sets of "num" anchors.

Yeah same, and I can't seem to find what's going on. Resorted to saving figs in a temp dir but this is super dirty. > AttributeError: module '_tkinter' has no...