HDR-IRSTD: Detection-Driven HDR Infrared Image Enhancement and Small Target Detection Based on HDR Infrared Image Enhancement

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The paper proposes a detection-driven framework that integrates high dynamic range (HDR) infrared image enhancement with small target detection. Unlike conventional low dynamic range (LDR)-based methods that often lose detail or distort targets during enhancement. Overall the paper is well written and the contribution is of clear value.

1.The paper generates HDR datasets (NUDT-SIRST-16bit, SIRST-16bit) by stretching existing LDR datasets. How well do these synthetic HDR images capture the complexity, sensor noise, and environmental variability of real HDR infrared imagery?

2. some references about DL image detection/segmentation to enhance the litterature review: Accurate identification and measurement of the precipitate area by two-stage deep neural networks in novel chromium-based alloys; Generative text-guided 3d vision-language pretraining for unified medical image segmentation

3. While the method claims a balance between efficiency and detection performance, AVM-UNet introduces extra complexity. How does it perform on resource-limited platforms (e.g., drones, satellites) compared to lightweight networks with fewer parameters?

4.Deep cooperative training strategies may improve metrics, but how interpretable are the learned enhancements and detections?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

1:.Refine the method description details and supplement the implementation details of key modules.

2.This ablation experiment is insufficient to demonstrate the overall validity of the model, and the lack of experimental parameter configuration details fails to establish the fairness of the comparative experiments.

3.Lack of edge device verification, with no performance testing or compatibility analysis conducted for edge embedded platforms.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

the paper has been greatly improved

Reviewer 2 Report

Comments and Suggestions for Authors

Following the author's careful revisions, the article features a well-designed experimental setup, with data sufficiently supporting the proposed method's validity and demonstrating substantial effort. It is recommended for acceptance in its current form.