DetR-ResNet50 trained on VoC

Files

• detr-b.pt: raw PyTorch checkpoint from this training run (~476 MB)
• class_wise_ap.csv: per-class average precision for each IoU threshold from 0.50 to 0.95
• cls_wise_ap_overall.png: plot of class-wise AP across the VOC classes (600x400)
• map_per_iou.csv: mean average precision values for IoU thresholds from 0.50 to 0.95
• map_per_iou.png: plot of mAP versus IoU threshold (600x400)