search

skumra / robotic-grasping

Antipodal Robotic Grasping using GR-ConvNet. IROS 2020.
Other
456 stars 119 forks source link

About trained_model evaluate #22

Closed catisthebest closed 3 years ago

catisthebest commented 3 years ago

Dear skumra:

thanks for your amazing work. i have a problem when i run your evaluate.py file, hope u can help me. thanks a lot.

when i run the evaluate.py , i have an abnormal IOU result : IOU Results: 4/5449 = 0.000734 the command i use , i put below: python evaluate.py --network trained-models/cornell-randsplit-rgbd-grconvnet3-drop1-ch32/epoch_19_iou_0.98 --dataset jacquard --dataset-path ../Jacquard --iou-eval --jacquard-output i don't know the reason that why the IOU results is so low..
i did not change the model and code . could u help me , thank you!

catisthebest commented 3 years ago

i think the reason is the model is trained on cornell dataset ... - - i try to train model on jacquard dataset today . It seems like the model training speed on jacquard is much slower than training on cornell dataset. ？？

skumra commented 3 years ago

The two datasets are a lot different from each other and that's the reason for the low IOU results. The cornell dataset is on real objects as compared to the simulated objects in jacquard dataset.

Moreover, the training time using the jacquard dataset is slower because it is much larger than the cornell dataset (8k vs 1.1 million grasps).

catisthebest commented 3 years ago

got it . thank you~