assert len(roidb) > 0

[chakpongchung]

I am trying to train on my own dataset. I am following your "Training on KITTI detection dataset

" to convert kikki to pascal VOC and then run the training. What could go wrong here? It seems like the code cant load the roidb from

roidb = get_training_roidb(imdb)

This is what my data looks like:

microway:data$ pwd /data/home/microway/experiments/TFFRCNN/data microway:KITTI$ tree . ├── testing │   ├── image_2 │   │   ├── 0000000000.png │   │   ├── 0000000001.png │   │   ├── 0000000002.png │   │   ├── 0000000003.png │   │   ├── 0000000004.png │   │   ├── 0000000005.png │   │   ├── 0000000006.png │   │   ├── 0000000007.png │   │   ├── 0000000008.png │   │   └── 0000000009.png │   └── label_2 │   ├── 0000000000.txt │   ├── 0000000001.txt │   ├── 0000000002.txt │   ├── 0000000003.txt │   ├── 0000000004.txt │   ├── 0000000005.txt │   ├── 0000000006.txt │   ├── 0000000007.txt │   ├── 0000000008.txt │   └── 0000000009.txt └── training ├── image_2 │   ├── 0000000000.png │   ├── 0000000001.png │   ├── 0000000002.png │   ├── 0000000003.png │   ├── 0000000004.png │   ├── 0000000005.png │   ├── 0000000006.png │   ├── 0000000007.png │   ├── 0000000008.png │   └── 0000000009.png └── label_2 ├── 0000000000.txt ├── 0000000001.txt ├── 0000000002.txt ├── 0000000003.txt ├── 0000000004.txt ├── 0000000005.txt ├── 0000000006.txt ├── 0000000007.txt ├── 0000000008.txt └── 0000000009.txt

python /$TFFRCNN/experiments/scripts/kitti2pascalvoc.py --kitti $TFFRCNN/data/KITTI --out $TFFRCNN/data/KITTIVOC

microway:data$ tree KITTIVOC/ KITTIVOC/ ├── Annotations │   ├── 0000000000.xml │   ├── 0000000001.xml │   ├── 0000000002.xml │   ├── 0000000003.xml │   ├── 0000000004.xml │   ├── 0000000005.xml │   ├── 0000000006.xml │   ├── 0000000007.xml │   ├── 0000000008.xml │   └── 0000000009.xml ├── ImageSets │   ├── Layout │   ├── Main │   │   ├── car_train.txt │   │   ├── car_trainval.txt │   │   ├── car_val.txt │   │   ├── cyclist_train.txt │   │   ├── cyclist_trainval.txt │   │   ├── cyclist_val.txt │   │   ├── dontcare_train.txt │   │   ├── dontcare_trainval.txt │   │   ├── dontcare_val.txt │   │   ├── pedestrian_train.txt │   │   ├── pedestrian_trainval.txt │   │   ├── pedestrian_val.txt │   │   ├── train.txt │   │   ├── trainval.txt │   │   └── val.txt │   └── Segmentation ├── JPEGImages │   ├── 0000000000.jpg │   ├── 0000000001.jpg │   ├── 0000000002.jpg │   ├── 0000000003.jpg │   ├── 0000000004.jpg │   ├── 0000000005.jpg │   ├── 0000000006.jpg │   ├── 0000000007.jpg │   ├── 0000000008.jpg │   └── 0000000009.jpg ├── SegmentationClass └── SegmentationObject

python ./faster_rcnn/train_net.py
--gpu 0
--weights ./data/pretrain_model/VGG_imagenet.npy
--imdb kittivoc_train
--iters 160000
--cfg ./experiments/cfgs/faster_rcnn_kitti.yml
--network VGGnet_train

Traceback (most recent call last): File "./faster_rcnn/train_net.py", line 110, in restore=bool(int(args.restore))) File "./faster_rcnn/../lib/fast_rcnn/train.py", line 398, in train_net sw = SolverWrapper(sess, network, imdb, roidb, output_dir, logdir= log_dir, pretrained_model=pretrained_model) File "./faster_rcnn/../lib/fast_rcnn/train.py", line 44, in init self.bbox_means, self.bbox_stds = rdl_roidb.add_bbox_regression_targets(roidb) File "./faster_rcnn/../lib/roi_data_layer/roidb.py", line 57, in add_bbox_regression_targets assert len(roidb) > 0 AssertionError