SEFormer: Structure Embedding Transformer for 3D Object Detection

Authors: Xiaoyu Feng, Heming Du, Hehe Fan, Yueqi Duan, Yongpan Liu

AAAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments show that the proposed architecture can achieve SOTA results on the Waymo Open Dataset, one of the most significant 3D detection benchmarks for autonomous driving.
Researcher Affiliation Academia 1Tsinghua University, 2Australian National University, 3National University of Singapore
Pseudocode No The paper does not contain a clearly labeled pseudocode or algorithm block.
Open Source Code Yes https://github.com/tdzdog/SEFormer.
Open Datasets Yes The Waymo dataset contains 1000 Li DAR sequences in total. These sequences are further split into 798 training sequences (including around 158k Li DAR samples) and 202 validation sequences (including around 40k Li DAR samples).
Dataset Splits Yes The Waymo dataset contains 1000 Li DAR sequences in total. These sequences are further split into 798 training sequences (including around 158k Li DAR samples) and 202 validation sequences (including around 40k Li DAR samples).
Hardware Specification Yes We use 4 RTX 3090 GPUs to train the entire network with batch size 8.
Software Dependencies No The paper mentions using "Adam W optimizer" and "one-cycle policy" but does not specify software names with version numbers (e.g., Python, PyTorch, TensorFlow versions).
Experiment Setup Yes We use 4 RTX 3090 GPUs to train the entire network with batch size 8. We adopt Adam W optimizer and one-cycle policy(Smith and Topin 2019) with division factor 10 and momentum ranges from 0.95 to 0.85 to train the model. The learning rate is initialized with 0.003. The training time is 40 epochs.