Very low confidance score after 10,000 iterations

[ztianlin]

Hi, I have modified your code to train an EfficientNet-SSD. After the first 10,000 iterations (loss converged to 7), I picked up the model and evaluated it on COCO2017. I got really bad results as fellow. Is it necessary to train 400,000 iterations? Cause doing that almost costs me 100 hours with four NVIDIA Tesla V100 gpus.

 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.004
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.012
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.002
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.006
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.008
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.018
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.042
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.050
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.008
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.055
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.079

[Damon2019]

@ztianlin hi,can you tell me how to evaluate COCO dataset,thanks

[leiyaohui]

@ztianlin I have the same problem. Have you solved it？

[bspivey]

After 50,000 iterations, my model was able to predict three people with < 80% confidence instead of the 7 people in the demo file. Curious if anyone has found more optimal settings?

[szy4017]

Hi, I have modified your code to train an EfficientNet-SSD. After the first 10,000 iterations (loss converged to 7), I picked up the model and evaluated it on COCO2017. I got really bad results as fellow. Is it necessary to train 400,000 iterations? Cause doing that almost costs me 100 hours with four NVIDIA Tesla V100 gpus.

 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.004
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.012
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.002
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.006
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.008
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.018
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.042
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.050
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.008
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.055
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.079

i have the same problem. help!