yrj90 / mil-pain

Code and notes for experiments of mil-pain

mil-pain

Code and notes for experiments of mil-pain

mil-pain's Issues

pytorch

RuntimeError: cuda runtime error (59) : device-side assert triggered at /opt/conda/conda-bld/pytorch_1533672544752/work/aten/src/THC/generic/THCTensorCopy.cpp:70

/opt/conda/conda-bld/pytorch_1533672544752/work/aten/src/THCUNN/ClassNLLCriterion.cu:105: void cunn_ClassNLLCriterion_updateOutput_kernel(Dtype *, Dtype *, Dtype *, long *, Dtype *, int, int, int, int, long) [with Dtype = float, Acctype = float]: block: [0,0,0], thread: [1,0,0] Assertion t >= 0 && t < n_classes failed.
THCudaCheck FAIL file=/opt/conda/conda-bld/pytorch_1533672544752/work/aten/src/THC/generic/THCTensorCopy.cpp line=70 error=59 : device-side assert triggered

Steps of train mil for pain datasets (modification of data and structure)

MIL part is from RstarCNN.

1. Prepare data into voc form and make a symbolic link to the data folder
2. In ROOT/lib/datasets/'name-of-db'.py and a 'name-of-db-eval.py' corresponding the pascal_voc.py and voc_eval.py
3. In the ROOT/lib/datasets/factory.py , add a new entry for your own dataset.
4. In ROOT/models/, change an existing one VGG16_RstarCNN. Remember to change the path within the solver and the number of classes in the train and test prototxt.
5. ? how to block the primary part of RstarCNN, and only use the mil part?
6. Change corresponding codes in ./tools/train_net.py, and ./tools/test_net.py