yolo3 INT8 conversion drops mAP incredibly #263

New Issue

2021-09-05T16:05:55+02:00

MohamedElsaeidy commented

2021-09-05 16:05:55 +02:00

(Migrated from github.com)

i changed conf_treshhold from 0.3 to 0.001 and tried calibration with 4951 images and still mAP is very bad compared to int8 of yolo4 , the mAP just increased for all models because of conf_tresh except that nothing changed and still bad for yolo3 int8. any suggestion or explanation for what causing that would be greatful. thanks
note : all other conversions for fp32 16 int8 for yolo3 and yolo4 are good and very close to published results

Classes: 80 mAP 0.5:	0.27588
Classes: 80 mAP 0.55:	0.249992
Classes: 80 mAP 0.6:	0.216834
Classes: 80 mAP 0.65:	0.17984
Classes: 80 mAP 0.7:	0.137224
Classes: 80 mAP 0.75:	0.0899831
Classes: 80 mAP 0.8:	0.048967
Classes: 80 mAP 0.85:	0.0190462
Classes: 80 mAP 0.9:	0.00424098
Classes: 80 mAP 0.95:	0.000314722
mAP 0.5:0.95 = 0.122232
avg precision: 0.119544	avg recall: 0.523164	avg f1 score:0.194617

i changed conf_treshhold from 0.3 to 0.001 and tried calibration with 4951 images and still mAP is very bad compared to int8 of yolo4 , the mAP just increased for all models because of conf_tresh except that nothing changed and still bad for yolo3 int8. any suggestion or explanation for what causing that would be greatful. thanks note : all other conversions for fp32 16 int8 for yolo3 and yolo4 are good and very close to published results <pre>Classes: 80 mAP 0.5: 0.27588 Classes: 80 mAP 0.55: 0.249992 Classes: 80 mAP 0.6: 0.216834 Classes: 80 mAP 0.65: 0.17984 Classes: 80 mAP 0.7: 0.137224 Classes: 80 mAP 0.75: 0.0899831 Classes: 80 mAP 0.8: 0.048967 Classes: 80 mAP 0.85: 0.0190462 Classes: 80 mAP 0.9: 0.00424098 Classes: 80 mAP 0.95: 0.000314722 mAP 0.5:0.95 = 0.122232 avg precision: 0.119544 avg recall: 0.523164 avg f1 score:0.194617 </pre>

👍 1

mive93 commented

2021-11-25 11:10:57 +01:00

(Migrated from github.com)

Hi @MohamedElsaeidy,
It is normal a drop of performance when going from float (even half precision) to int8. I do agree that the jump is very big. Here I can attach a chart with several comparison we did for a journal paper (still under review), in which you can find a lor of information about different platforms and network for all the data types.
Hope it helps.

Hi @MohamedElsaeidy, It is normal a drop of performance when going from float (even half precision) to int8. I do agree that the jump is very big. Here I can attach a chart with several comparison we did for a journal paper (still under review), in which you can find a lor of information about different platforms and network for all the data types. Hope it helps. ![image](https://user-images.githubusercontent.com/11939259/143421929-a32d0d37-c00d-41b0-814c-84ab849d969e.png)

This repo is archived. You cannot comment on issues.

1 Participants

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: mmr/tkDNN#263