rtdetr-v2-r50-finetune-9

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: 6.4316
• Map: 0.49
• Map 50: 0.781
• Map 75: 0.5504
• Map Small: 0.4762
• Map Medium: 0.6106
• Map Large: -1.0
• Mar 1: 0.3304
• Mar 10: 0.6631
• Mar 100: 0.69
• Mar Small: 0.6653
• Mar Medium: 0.7529
• Mar Large: -1.0
• Map Artemia: 0.49
• Mar 100 Artemia: 0.69

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	250	13.7176	0.2126	0.3985	0.1975	0.1372	0.4217	-1.0	0.2682	0.5863	0.6452	0.5554	0.7647	-1.0	0.2126	0.6452
243.12	2.0	500	8.2150	0.3694	0.6877	0.3618	0.297	0.4979	-1.0	0.3318	0.6031	0.6383	0.5962	0.6964	-1.0	0.3694	0.6383
243.12	3.0	750	8.2831	0.4746	0.8355	0.4849	0.3978	0.583	-1.0	0.3723	0.595	0.6352	0.5891	0.6964	-1.0	0.4746	0.6352
14.0162	4.0	1000	8.2865	0.1154	0.2137	0.108	0.3503	0.1702	-1.0	0.3536	0.5819	0.5885	0.5315	0.6655	-1.0	0.1154	0.5885
14.0162	5.0	1250	8.1979	0.3304	0.6291	0.2988	0.3337	0.4718	-1.0	0.3526	0.581	0.5988	0.5435	0.6734	-1.0	0.3304	0.5988
12.2257	6.0	1500	8.0874	0.4577	0.8447	0.4189	0.3797	0.5843	-1.0	0.3642	0.5623	0.5685	0.4897	0.6748	-1.0	0.4577	0.5685
12.2257	7.0	1750	8.4536	0.4003	0.801	0.3266	0.3084	0.561	-1.0	0.3336	0.5645	0.5776	0.5223	0.6532	-1.0	0.4003	0.5776
11.2516	8.0	2000	8.1441	0.4345	0.8414	0.4042	0.3667	0.5755	-1.0	0.3645	0.5698	0.5735	0.519	0.6482	-1.0	0.4345	0.5735
11.2516	9.0	2250	8.9297	0.3797	0.7669	0.3187	0.2963	0.5666	-1.0	0.3153	0.5598	0.5645	0.5038	0.6468	-1.0	0.3797	0.5645
10.2227	10.0	2500	8.4697	0.4315	0.8204	0.3933	0.3543	0.572	-1.0	0.3564	0.5639	0.5657	0.5049	0.6489	-1.0	0.4315	0.5657
10.2227	11.0	2750	8.8899	0.3787	0.7271	0.3401	0.3142	0.5382	-1.0	0.3364	0.571	0.5773	0.5098	0.6691	-1.0	0.3787	0.5773
9.3841	12.0	3000	8.7802	0.3904	0.7519	0.3354	0.3059	0.5668	-1.0	0.3234	0.5601	0.5601	0.488	0.6576	-1.0	0.3904	0.5601
9.3841	13.0	3250	9.2236	0.3568	0.7077	0.306	0.2682	0.563	-1.0	0.3097	0.5583	0.5598	0.4859	0.6604	-1.0	0.3568	0.5598
8.5284	14.0	3500	9.0471	0.3744	0.7302	0.3176	0.2931	0.5665	-1.0	0.3125	0.562	0.562	0.4929	0.6561	-1.0	0.3744	0.562
8.5284	15.0	3750	9.1800	0.3691	0.7034	0.336	0.2879	0.5633	-1.0	0.3031	0.5536	0.5536	0.4842	0.6482	-1.0	0.3691	0.5536
7.8078	16.0	4000	8.9976	0.4009	0.7641	0.351	0.3165	0.5737	-1.0	0.3274	0.5505	0.5505	0.4734	0.6547	-1.0	0.4009	0.5505
7.8078	17.0	4250	9.6222	0.3478	0.6749	0.3029	0.2589	0.575	-1.0	0.2988	0.5573	0.5583	0.4875	0.654	-1.0	0.3478	0.5583
7.0036	18.0	4500	9.3709	0.3694	0.7027	0.3222	0.2817	0.5676	-1.0	0.3093	0.5514	0.5523	0.4783	0.6532	-1.0	0.3694	0.5523
7.0036	19.0	4750	9.6374	0.3628	0.6967	0.3224	0.2695	0.574	-1.0	0.2969	0.5545	0.5561	0.4853	0.6525	-1.0	0.3628	0.5561
6.3958	20.0	5000	9.7867	0.3555	0.684	0.3096	0.2599	0.5612	-1.0	0.2941	0.5539	0.5545	0.4821	0.6532	-1.0	0.3555	0.5545

Framework versions

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