search

tianzhi0549 / FCOS

FCOS: Fully Convolutional One-Stage Object Detection (ICCV'19)
https://arxiv.org/abs/1904.01355
Other
3.27k stars 629 forks source link

A problem during inference on my own datasets #133

Open stellating opened 5 years ago

stellating commented 5 years ago

I have trained FCOS on my own dataset which has 3 classes, and I'v got a model with 87500 iterations. But when I try to take an inference on my own datasets, I get the following massage:

Running per image evaluation... Evaluate annotation type bbox DONE (t=0.62s). Accumulating evaluation results... DONE (t=0.22s). Average Precision (AP) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.000 Average Precision (AP) @[ IoU=0.50 | area= all | maxDets=100 ] = 0.000 Average Precision (AP) @[ IoU=0.75 | area= all | maxDets=100 ] = 0.000 Average Precision (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000 Average Precision (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = -1.000 Average Precision (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000 Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets= 1 ] = 0.000 Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets= 10 ] = 0.000 Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.000 Average Recall (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000 Average Recall (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = -1.000 Average Recall (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000 Maximum f-measures for classes: [-0.0, -0.0, 0.0] Score thresholds for classes (used in demos for visualization purposes): [-1.0, -1.0, 0.9576103687286377] 2019-09-06 10:20:00,552 fcos_core.inference INFO: OrderedDict([('bbox', OrderedDict([('AP', 0.0), ('AP50', 0.0), ('AP75', 0.0), ('APs', -1.0), ('APm', -1.0), ('APl', 0.0)]))])

Could you please help me?

tianzhi0549 commented 5 years ago

@stellating I suggest you visualize your detection results to ensure your model works well.

dreamhighchina commented 5 years ago

@stellating the same with you，how do you deal with it？

Finniu commented 4 years ago

same as me

liudanyang commented 4 years ago

same as me

I got the same problem. Did you fix it?

Finniu commented 4 years ago

same as me

I got the same problem. Did you fix it?

no, I think it is the problem that the num of training class is 81 in source code, but mine is different, and he didn't say how to change it. When i was training, i just loaded all backbone weights, but still doesn't work, XD. So i decide to build a trainer by myself

liudanyang commented 4 years ago

no, I think it is the problem that the num of training class is 81 in source code, but mine is different, and he didn't say how to change it. When i was training, i just loaded all backbone weights, but still doesn't work, XD. So i decide to build a trainer by myself

Well, you can change the category number in fcos_core/config/default.py. But...I still didn't figure out why this issue happened.

liudanyang commented 4 years ago

oh...I found the problem..the fetch_data function that I wrote has a bug....

JerryIndus commented 4 years ago

@liudanyang I met similar problem, and could you tell me the how wrong about fetch_data function detailly? Maybe, i can learn from it, thank you.

punknownq commented 4 years ago

@JerryIndus Hi.Have you solved the problem?

punknownq commented 4 years ago

@liudanyang Could you tell me how do you fix the problem?

pp00704831 commented 3 years ago

Hello, Does anyone fix it?

XDWang97 commented 2 years ago

Does anyone fix it?