- Rating
- 3(+1)
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
This paper addresses the exposure bias of diffusion-based detection transformers and proposes self-trajectory denoising. The denoising trajectory starts from the top-k encoder proposals rather than random noise, and the noise is added in logit space, which bounds the perturbation. During training, each sample follows two denoising steps along the trajectory, where the second step takes a re-noised copy of the first step's prediction as input and gradients are stopped between the two steps. Ablations quantify the respective contributions of the encoder prior, dense augmentation, and the matchability-aware loss, and show that accuracy varies very little across different numbers of denoising steps.
Critical strengths
The paper identifies the exposure bias of diffusion-based detection transformers and proposes a solution: the denoising trajectory starts from encoder proposals shared by training and inference, and a second training step trains the decoder on re-noised copies of its own predictions, thereby providing an informative starting point for denoising. With a short training schedule, PriorDiffuDINO clearly surpasses DiffuDINO trained equally long and approaches its version trained several times longer. The gains appear exactly where the theoretical analysis predicts, which makes the improvements mechanistically convincing.
Critical weaknesses
- The appendix shows that a Gaussian start with more steps nearly matches the prior-based start, so the encoder prior may act mainly as an inference-efficiency device rather than an accuracy device. However, the paper reports no latency, throughput, or FLOPs for either training or inference.
- The two-step training design is not ablated. The paper does not justify why two training steps suffice, nor does it ablate the number of training steps or the maximum noise level.
- The matchability-aware loss and dense augmentation are borrowed from DEIM, yet DEIM itself is not included in the comparison.
Questions for authors
- Could you report the training time, inference latency, or FLOPs of PriorDiffuDINO, DINO, and DiffuDINO under different numbers of steps?
- Why does Figure 2 mark the steps as Step ≥ 2? Why using two training steps? Have you ablated the number of training steps or the maximum noise level?
- Could you include DEIM in the comparison?
- Could you provide more hyperparameter sensitivity analyses?
- Could you quantify the contribution of FiLM conditioning?