Code for the CubeRefine R-CNN model of our CVPRW '23 paper "Parcel3D: Shape Reconstruction From Single RGB Images for Applications in Transportation Logistics".
TAMPAR: Tampering detection for parcel logistics from our WACV '24 paper "TAMPAR: Visual Tampering Detection for Parcel Logistics in Postal Supply Chains"
CubeRefine R-CNN: architecture for 3D Reconstruction for potentially damaged parcels from our CVPRW '23 paper "Parcel3D: Shape Reconstruction From Single RGB Images for Applications in Transportation Logistics"
lanelet2anchors: pip package for the generation of diverse map-based anchor paths from our CVPRW' 23 paper "Lanelet2 for nuScenes: Enabling Spatial Semantic Relationships and Diverse Map-based Anchor Paths"
Repos of paper "Scrape, Cut, Paste and Learn: Automated Dataset Generation Applied to Parcel Logistics"
My code runs properly for training process (upto 1000 iteration) but stuck at validation process. I am using 4 GPUS to run the code.
Stuck after evaluating following lines:
[08/10 02:48:12] d2.evaluation.evaluator INFO: Start inference on 1 batches
[08/10 02:49:00] d2.evaluation.evaluator INFO: Inference done 1/1. Dataloading: 0.0000 s/iter. Inference: 0.1117 s/iter. Eval: 0.0061 s/iter. Total: 0.1178 s/iter. ETA=0:00:00
[08/10 02:49:06] d2.evaluation.evaluator INFO: Total inference time: 0:00:05.186749 (5.186749 s / iter per device, on 4 devices)
[08/10 02:49:06] d2.evaluation.evaluator INFO: Total inference pure compute time: 0:00:00 (0.111703 s / iter per device, on 4 devices)
Also, could you please provide output visualization and test script separately for checking the performance on my dataset.
Thank you in advance for your cooperation.