Implementation of paper - YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors
GNU General Public License v3.0
13.41k
stars
4.23k
forks
source link
Reproducing results from paper #2025
Open
LauraBraad opened 7 months ago
I am trying to reproduce the results from the paper, specifically :
Average Precision (AP) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.51206 Average Precision (AP) @[ IoU=0.50 | area= all | maxDets=100 ] = 0.69730 Average Precision (AP) @[ IoU=0.75 | area= all | maxDets=100 ] = 0.55521
I am running the line:
'python -m torch.distributed.launch --nproc_per_node 2 --master_port 9527 train.py --workers 8 --device 0,1 --sync-bn --batch-size 16 --epochs 600 --data data/coco.yaml --img 640 640 --cfg cfg/training/yolov7.yaml --weights '' --name yolov7_retrained --hyp data/hyp.scratch.p5.yaml'
Where I have changed the batch size and epoch so I was able to run on GPU. But I get way worse results:
Average Precision (AP) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.339 Average Precision (AP) @[ IoU=0.50 | area= all | maxDets=100 ] = 0.495 Average Precision (AP) @[ IoU=0.75 | area= all | maxDets=100 ] = 0.367
What is the issue here?