Code release for our NeurIPS 2023 paper "Uni3DETR: Unified 3D Detection Transformer", our ECCV 2024 paper "OV-Uni3DETR: Towards Unified Open-Vocabulary 3D Object Detection via Cycle-Modality Propagation"
Apache License 2.0
81
stars
3
forks
source link
Hyperparameters provided in the config do not seem to match those provided by the implementation details in the paper #7
For example, the config trains scannet_large for 40 epochs, but the paper says the model is trained for 240 epochs.
Can the hyperparameters provided in the config be used to reproduce the results in the paper?
Or do I have to modify them based on the paper? But if that is the case, details like what weights to use for each component of the loss are also missing.
For example, the config trains scannet_large for 40 epochs, but the paper says the model is trained for 240 epochs.
Can the hyperparameters provided in the config be used to reproduce the results in the paper?
Or do I have to modify them based on the paper? But if that is the case, details like what weights to use for each component of the loss are also missing.