Closed edwardnguyen1705 closed 4 years ago
I think dataset.coco
means the object of COCO
. which is your loaded coco format dataset. it can be considerate as ground truth coco for evaluation.
for CustomDataset evaluation got mAP=0, check the annotation iscrowd enable or not, if you using RLE for representation of segmentation mask, you should enable iscrowd=1 when calculate mask iou
@jinfagang Thanks for sharing.
for CustomDataset evaluation got mAP=0, check the annotation iscrowd enable or not, if you using RLE for representation of segmentation mask, you should enable iscrowd=1 when calculate mask iou
I do not quite understand this one. Yes, my dataset uses RLE for representation of segmentation mask.
@edwardnguyen1705 coco loadRes will force iscrowd to false. You can take a close look at coco evaluation code.
@jinfagang , Thank you for your response.
@edwardnguyen1705 ,how to modify test_ins.py when I use own data?thanks
@haibucuo0 you can take a look at my fork: solo
Dear @WXinlong ,
After training on a custom dataset (inherited from the
CocoDataset
, just like the way in thecityscapes.py
), the visualization looks great, but mAP = 0. I see in thetest_insts.py
coco_eval(result_file, eval_types, dataset.coco)
Is it the right way?