DETR VAL mAP 0.0000

[FPPMXB]

MMdetection3.0 Problem: AP values on the DETR validation set are all 0.0000

Conditions:

1. NWPU-VHR-10 data set was selected, including 455 training sets and 195 verification and testing sets.

2. It can be successfully trained on both Faster-rcnn and YOLOv3 models, but during DETR training, the verification index is all 0.0000.

3. Pre-training model is used

Question:

1. All AP values on DETR verification set are 0.0000

2. In addition, it is also a problem when training Co-DETR, that is to say, as long as the DETR based model has this problem.

Is it because the data set is too small? Or do you need to pay special attention to the adjustment of some specific parameters?

[FPPMXB]

[FPPMXB]

As shown in the figure above, it is always 0.0000, even after dozens of epochs have not changed.

[kennedyk1]

I had the same problem in my custom dataset with 1100 images (800 train, 150 test, 150 validation) on DETR, and successfully on YoloX.

[FPPMXB]

可能是数据量太小的原因，继续transformer的模型太吃数据了，你可以看csdn这篇文章http://t.csdnimg.cn/hvKdq 。我做了一些验证性的实验做了一些总结，希望对你有帮助

---Original--- From: @.> Date: Tue, Dec 12, 2023 23:05 PM To: @.>; Cc: @.@.>; Subject: Re: [open-mmlab/mmdetection] DETR VAL mAP 0.0000 (Issue #11262)

I had the same problem in my custom dataset with 1100 images (800 train, 150 test, 150 validation) on DETR, and successfully on YoloX.

image.png (view on web)

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

[neurosynapse]

Same issue. After doing inference with these checkpoints no detection can be seen. Tried most of the configs available for this model. Tried locally and in cloud. Same results.

[hysjly]

When I was training a custom dataset with Mask R-CNN, I encountered the same problem. Has your issue been resolved?（The dataset I used contains 3,000 images.）

[skywalkerfmc]

Same issue

[sjxxrh]

Same issue

[Goddaman]

same issue, did u fix it?

[hysjly]

My issue at the time was that the annotations in the JSON file were incorrect. I hope this is helpful to you.

hysjly @.***

---Original--- From: @.> Date: Tue, Nov 11, 2025 22:25 PM To: @.>; Cc: @.@.>; Subject: Re: [open-mmlab/mmdetection] DETR VAL mAP 0.0000 (Issue #11262)

Goddaman left a comment (open-mmlab/mmdetection#11262)

same issue, did u fix it?

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>