rtdetr-v2-r50-finetune-10

This model is a fine-tuned version of PekingU/rtdetr_v2_r50vd on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 6.5959
• Map: 0.5657
• Map 50: 0.9031
• Map 75: 0.6448
• Map Small: 0.542
• Map Medium: 0.6198
• Map Large: -1.0
• Mar 1: 0.3122
• Mar 10: 0.6657
• Mar 100: 0.6972
• Mar Small: 0.6786
• Mar Medium: 0.7312
• Mar Large: -1.0
• Map Artemia: 0.5657
• Mar 100 Artemia: 0.6972

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 300
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Map	Map 50	Map 75	Map Small	Map Medium	Map Large	Mar 1	Mar 10	Mar 100	Mar Small	Mar Medium	Mar Large	Map Artemia	Mar 100 Artemia
No log	1.0	160	16.9213	0.2329	0.4481	0.2146	0.1534	0.4298	-1.0	0.2812	0.5971	0.6517	0.578	0.7688	-1.0	0.2329	0.6517
No log	2.0	320	10.6024	0.0663	0.1225	0.0611	0.3307	0.2671	-1.0	0.3053	0.5121	0.6092	0.5669	0.6762	-1.0	0.0663	0.6092
No log	3.0	480	8.6030	0.4819	0.8201	0.5505	0.4192	0.5988	-1.0	0.3855	0.587	0.6266	0.5677	0.72	-1.0	0.4819	0.6266
22.7285	4.0	640	8.2806	0.4813	0.8429	0.4975	0.4122	0.6038	-1.0	0.3821	0.599	0.6227	0.5638	0.7163	-1.0	0.4813	0.6227
22.7285	5.0	800	8.7470	0.4506	0.8273	0.4127	0.3734	0.573	-1.0	0.371	0.5691	0.6295	0.5772	0.7125	-1.0	0.4506	0.6295
22.7285	6.0	960	8.1180	0.4739	0.8491	0.5071	0.4083	0.577	-1.0	0.3995	0.5768	0.6092	0.5598	0.6875	-1.0	0.4739	0.6092
13.426	7.0	1120	8.4592	0.4793	0.836	0.4908	0.4152	0.582	-1.0	0.3947	0.5614	0.5826	0.5268	0.6712	-1.0	0.4793	0.5826
13.426	8.0	1280	8.7920	0.432	0.8077	0.4043	0.372	0.5553	-1.0	0.3681	0.5527	0.5734	0.5252	0.65	-1.0	0.432	0.5734
13.426	9.0	1440	8.6275	0.4646	0.8242	0.4859	0.3929	0.5831	-1.0	0.387	0.5667	0.5773	0.5173	0.6725	-1.0	0.4646	0.5773
11.468	10.0	1600	9.4757	0.4087	0.7521	0.3975	0.3279	0.5527	-1.0	0.3676	0.5599	0.5841	0.5402	0.6538	-1.0	0.4087	0.5841
11.468	11.0	1760	8.6502	0.4373	0.8327	0.3842	0.3706	0.5516	-1.0	0.3725	0.5551	0.5749	0.5165	0.6675	-1.0	0.4373	0.5749
11.468	12.0	1920	8.6194	0.4475	0.8151	0.4207	0.3825	0.5657	-1.0	0.3754	0.5502	0.558	0.4953	0.6575	-1.0	0.4475	0.558
9.9247	13.0	2080	8.8678	0.459	0.8499	0.4322	0.3966	0.5656	-1.0	0.3855	0.5565	0.558	0.5008	0.6488	-1.0	0.459	0.558
9.9247	14.0	2240	8.9954	0.4337	0.7967	0.3763	0.3799	0.5556	-1.0	0.3787	0.557	0.5589	0.5087	0.6388	-1.0	0.4337	0.5589
9.9247	15.0	2400	8.6554	0.4518	0.8197	0.4208	0.3802	0.5793	-1.0	0.3633	0.57	0.5749	0.515	0.67	-1.0	0.4518	0.5749
8.476	16.0	2560	8.8688	0.4445	0.8152	0.4067	0.3781	0.5593	-1.0	0.3676	0.5585	0.5609	0.5071	0.6463	-1.0	0.4445	0.5609
8.476	17.0	2720	8.9928	0.4462	0.8087	0.4346	0.3778	0.5681	-1.0	0.3739	0.5633	0.5662	0.511	0.6538	-1.0	0.4462	0.5662
8.476	18.0	2880	8.9247	0.4579	0.8288	0.4535	0.3923	0.5685	-1.0	0.3734	0.5594	0.5599	0.5024	0.6513	-1.0	0.4579	0.5599
7.3216	19.0	3040	8.9952	0.4469	0.821	0.4372	0.3844	0.5577	-1.0	0.3744	0.5589	0.5594	0.5087	0.64	-1.0	0.4469	0.5594
7.3216	20.0	3200	9.1037	0.4437	0.8142	0.4389	0.3781	0.5629	-1.0	0.3739	0.5604	0.5609	0.5071	0.6463	-1.0	0.4437	0.5609

Framework versions

• Transformers 4.57.1
• Pytorch 2.8.0+cu128
• Datasets 4.2.0
• Tokenizers 0.22.1