AttributeError: 'EasyDict' object has no attribute 'PYTILES_PATH'

[R1ncy]

 Thanks for you nice work :). 
 I run the infer.py on test dataset and the results are very good. But when I run train.py, the error is:

self.tiles = Tiles(self.cfg.TRAIN.PARALLEL_TILES, pytiles_path=cfg.DIR.PYTILES_PATH, AttributeError: 'EasyDict' object has no attribute 'PYTILES_PATH'

PYTILES_PATH is not in cfg file and the params seem mismatch with https://github.com/tansor/VecRoad/blob/master/utils/tileloader.py#L159

[tansor]

Hi R1ncy, I have corrected the parameters in the code, let's see if it works. (Sorry that I don't have the training environment right now, so there may still be some errors when I reformat codes.)

[R1ncy]

thank you, I will try it

[R1ncy]

Hi tansor, i run train.py and it reports AttributeError: 'EasyDict' object has no attribute 'TILE_DIR' https://github.com/tansor/VecRoad/blob/master/utils/OSMDataset.py#L27

is this cfg.IMAGERY_DIR? I change it to cfg.IMAGERY_DIR then FileNotFoundError: [Errno 2] No such file or directory: 'data/input/regions/pytiles.json' Am i missing this pytiles.json? Thanks

[tansor]

Try 'data/input/regions/all_regions.txt'. (Updated code)

[R1ncy]

Hi tansor, seems 'road_seg_thick3' is not defined in function 'make_path_input' https://github.com/tansor/VecRoad/blob/master/utils/model_utils.py#L704

AttributeError: 'EasyDict' object has no attribute 'road_seg_thick3' https://github.com/tansor/VecRoad/blob/master/utils/OSMDataset.py#L122

[tansor]

Oh, I deleted them when reformatting codes. That's a mistake. Updated code.

[R1ncy]

Hi tansor, I train the vecroad model on my dataset and run infer on it. The roads in my dataset are at low level, like residential roads. The train and infer settings are nearly same with the default.yml. I check my road segmentation and junction nms result and they look nice. In the iteration process of 'infer_anchor', I print the length of 'search_vertices ' and 'graph' of path. The len(search_vertices) is increasing but the graph doesnt change(about 200 lines in .graph) even for 10k iterations. Could you give me some advice please? Thank you a lot

[tansor]

There may exist some corner cases that cause a dead recursive search. I'm afraid you have to print logs and visualize segmentation per inference step to debug. The proposed config is not guaranteed to have high performance in every self-defined dataset.