rtdetr-v2-r18-finetune-21

This model is a fine-tuned version of PekingU/rtdetr_v2_r18vd on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 4.5295
• Map: 0.4956
• Map 50: 0.8184
• Map 75: 0.5773
• Map Small: 0.4674
• Map Medium: 0.5942
• Map Large: -1.0
• Mar 1: 0.3201
• Mar 10: 0.6324
• Mar 100: 0.6803
• Mar Small: 0.6495
• Mar Medium: 0.7586
• Mar Large: -1.0
• Map Artemia: 0.4956
• Mar 100 Artemia: 0.6803

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 OptimizerNames.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: 50

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	7.4122	0.2071	0.4241	0.1928	0.1721	0.3426	-1.0	0.2452	0.534	0.6075	0.5317	0.7131	-1.0	0.2071	0.6075
166.2908	2.0	500	5.3821	0.4169	0.7786	0.3668	0.3383	0.5664	-1.0	0.3458	0.5713	0.6421	0.5957	0.7066	-1.0	0.4169	0.6421
166.2908	3.0	750	5.2432	0.3771	0.697	0.3457	0.3341	0.5578	-1.0	0.3405	0.5994	0.671	0.6151	0.7496	-1.0	0.3771	0.671
8.7492	4.0	1000	5.2816	0.4116	0.7935	0.388	0.3298	0.5809	-1.0	0.3408	0.6031	0.6424	0.5892	0.7161	-1.0	0.4116	0.6424
8.7492	5.0	1250	5.3025	0.3923	0.7681	0.3558	0.3202	0.5827	-1.0	0.3308	0.5875	0.6461	0.5839	0.7321	-1.0	0.3923	0.6461
7.8939	6.0	1500	5.2114	0.3724	0.7354	0.3615	0.3139	0.5164	-1.0	0.3321	0.5866	0.6274	0.5715	0.7044	-1.0	0.3724	0.6274
7.8939	7.0	1750	5.1376	0.4101	0.8007	0.375	0.3233	0.5813	-1.0	0.3299	0.5922	0.6234	0.5597	0.7109	-1.0	0.4101	0.6234
7.4828	8.0	2000	5.4221	0.4029	0.7725	0.3851	0.3128	0.6002	-1.0	0.3259	0.5931	0.6081	0.5484	0.6905	-1.0	0.4029	0.6081
7.4828	9.0	2250	5.5368	0.361	0.6974	0.3311	0.2732	0.5799	-1.0	0.3103	0.59	0.6065	0.5473	0.6876	-1.0	0.361	0.6065
7.0146	10.0	2500	5.3941	0.3805	0.7323	0.3546	0.2903	0.5731	-1.0	0.3196	0.591	0.605	0.5425	0.6912	-1.0	0.3805	0.605
7.0146	11.0	2750	5.4312	0.3789	0.7373	0.3572	0.2913	0.5781	-1.0	0.3165	0.5903	0.6059	0.536	0.7022	-1.0	0.3789	0.6059
6.7107	12.0	3000	5.6089	0.343	0.6786	0.3052	0.2383	0.58	-1.0	0.2841	0.5891	0.6087	0.5253	0.7234	-1.0	0.343	0.6087
6.7107	13.0	3250	5.6256	0.3674	0.7134	0.3531	0.271	0.5795	-1.0	0.314	0.5791	0.5894	0.5188	0.6869	-1.0	0.3674	0.5894
6.35	14.0	3500	5.5853	0.3302	0.6763	0.3106	0.2563	0.5676	-1.0	0.314	0.5773	0.5882	0.5274	0.6723	-1.0	0.3302	0.5882
6.35	15.0	3750	5.6741	0.3623	0.6982	0.3359	0.2735	0.59	-1.0	0.3081	0.5801	0.5866	0.5145	0.6861	-1.0	0.3623	0.5866
6.1359	16.0	4000	5.8476	0.3421	0.6707	0.3152	0.2481	0.5791	-1.0	0.286	0.5763	0.5829	0.4952	0.7044	-1.0	0.3421	0.5829
6.1359	17.0	4250	5.7151	0.3455	0.6843	0.3209	0.261	0.5733	-1.0	0.2835	0.5738	0.5798	0.5054	0.681	-1.0	0.3455	0.5798
5.845	18.0	4500	5.7376	0.3776	0.7401	0.3411	0.2842	0.5841	-1.0	0.2997	0.5682	0.5707	0.4871	0.6861	-1.0	0.3776	0.5707
5.845	19.0	4750	5.9623	0.3621	0.7138	0.3408	0.2643	0.5839	-1.0	0.2975	0.5632	0.5676	0.486	0.6781	-1.0	0.3621	0.5676
5.6541	20.0	5000	5.8622	0.3606	0.6961	0.3395	0.2612	0.5831	-1.0	0.2916	0.5651	0.5679	0.4849	0.6825	-1.0	0.3606	0.5679
5.6541	21.0	5250	5.9901	0.3042	0.6003	0.2877	0.2265	0.5377	-1.0	0.2988	0.552	0.5558	0.472	0.6715	-1.0	0.3042	0.5558
5.458	22.0	5500	5.9900	0.3453	0.6713	0.3473	0.2522	0.5711	-1.0	0.2966	0.5564	0.557	0.4694	0.6781	-1.0	0.3453	0.557
5.458	23.0	5750	5.9923	0.3435	0.6697	0.3413	0.2541	0.573	-1.0	0.2966	0.5589	0.5629	0.4806	0.6759	-1.0	0.3435	0.5629
5.268	24.0	6000	6.0327	0.3713	0.725	0.3619	0.2741	0.5742	-1.0	0.3016	0.5561	0.5564	0.4785	0.6642	-1.0	0.3713	0.5564
5.268	25.0	6250	6.0313	0.377	0.7459	0.3584	0.2788	0.5789	-1.0	0.3084	0.5586	0.5598	0.4801	0.6693	-1.0	0.377	0.5598
5.1915	26.0	6500	6.0146	0.353	0.6888	0.3358	0.2551	0.56	-1.0	0.3022	0.5564	0.5579	0.4726	0.6745	-1.0	0.353	0.5579
5.1915	27.0	6750	6.3282	0.3406	0.672	0.3339	0.2373	0.5767	-1.0	0.2869	0.5489	0.5498	0.4608	0.673	-1.0	0.3406	0.5498
5.0189	28.0	7000	6.3310	0.3417	0.6589	0.3259	0.2379	0.5749	-1.0	0.2794	0.5492	0.5495	0.4656	0.665	-1.0	0.3417	0.5495
5.0189	29.0	7250	6.3149	0.366	0.7139	0.3584	0.2657	0.5706	-1.0	0.3059	0.5467	0.547	0.4656	0.6599	-1.0	0.366	0.547
4.9079	30.0	7500	6.3977	0.3425	0.6706	0.3192	0.2439	0.5602	-1.0	0.2928	0.5445	0.5452	0.4651	0.6562	-1.0	0.3425	0.5452
4.9079	31.0	7750	6.4237	0.3429	0.6704	0.321	0.2413	0.5726	-1.0	0.2966	0.5477	0.5483	0.4683	0.6591	-1.0	0.3429	0.5483
4.7392	32.0	8000	6.4146	0.3672	0.7186	0.3458	0.2718	0.5705	-1.0	0.3047	0.5436	0.5436	0.4608	0.6577	-1.0	0.3672	0.5436
4.7392	33.0	8250	6.7436	0.3317	0.6491	0.3061	0.2309	0.5705	-1.0	0.3009	0.5505	0.5514	0.4731	0.6599	-1.0	0.3317	0.5514
4.7081	34.0	8500	6.5395	0.3364	0.6486	0.3299	0.2298	0.5748	-1.0	0.3009	0.5514	0.552	0.4672	0.6693	-1.0	0.3364	0.552
4.7081	35.0	8750	6.5957	0.3322	0.6564	0.3176	0.2336	0.5666	-1.0	0.3072	0.548	0.5483	0.4699	0.6569	-1.0	0.3322	0.5483
4.5899	36.0	9000	6.6439	0.2871	0.571	0.2661	0.1981	0.5586	-1.0	0.2988	0.5405	0.5458	0.4694	0.6518	-1.0	0.2871	0.5458
4.5899	37.0	9250	6.6730	0.3367	0.6515	0.3108	0.2331	0.5699	-1.0	0.2994	0.5461	0.5461	0.4618	0.6628	-1.0	0.3367	0.5461
4.538	38.0	9500	6.7328	0.3343	0.6624	0.3044	0.2322	0.5707	-1.0	0.2988	0.547	0.5474	0.4634	0.6635	-1.0	0.3343	0.5474
4.538	39.0	9750	6.9003	0.3292	0.6391	0.3117	0.2285	0.5774	-1.0	0.2928	0.5514	0.5517	0.4667	0.6693	-1.0	0.3292	0.5517
4.4147	40.0	10000	6.8279	0.3286	0.6453	0.3154	0.2271	0.5737	-1.0	0.2975	0.5445	0.5452	0.4645	0.6569	-1.0	0.3286	0.5452
4.4147	41.0	10250	6.9192	0.3136	0.6039	0.3022	0.2124	0.5615	-1.0	0.2872	0.5411	0.5417	0.4581	0.6577	-1.0	0.3136	0.5417
4.3455	42.0	10500	6.8160	0.3365	0.663	0.314	0.2389	0.566	-1.0	0.2913	0.5389	0.5399	0.4586	0.6526	-1.0	0.3365	0.5399
4.3455	43.0	10750	6.8642	0.3322	0.6554	0.2996	0.231	0.5659	-1.0	0.3097	0.5414	0.5421	0.4608	0.6547	-1.0	0.3322	0.5421
4.2857	44.0	11000	6.8900	0.3219	0.6285	0.3039	0.2232	0.5576	-1.0	0.3037	0.5374	0.5377	0.4548	0.6526	-1.0	0.3219	0.5377
4.2857	45.0	11250	6.9974	0.3335	0.6558	0.3154	0.2324	0.572	-1.0	0.2953	0.5442	0.5445	0.4591	0.6628	-1.0	0.3335	0.5445
4.2483	46.0	11500	7.0422	0.3298	0.6405	0.3076	0.2232	0.5711	-1.0	0.2975	0.5424	0.543	0.4543	0.6657	-1.0	0.3298	0.543
4.2483	47.0	11750	7.1007	0.3269	0.6385	0.3076	0.2228	0.5696	-1.0	0.2935	0.5408	0.5411	0.4532	0.6628	-1.0	0.3269	0.5411
4.1284	48.0	12000	7.1250	0.3311	0.6512	0.3064	0.2317	0.5671	-1.0	0.2963	0.543	0.5439	0.4591	0.6613	-1.0	0.3311	0.5439
4.1284	49.0	12250	7.0028	0.3421	0.6749	0.3151	0.2415	0.5686	-1.0	0.3031	0.5399	0.5399	0.4554	0.6569	-1.0	0.3421	0.5399
4.1162	50.0	12500	6.9799	0.341	0.6701	0.3142	0.2407	0.5706	-1.0	0.3044	0.5371	0.5371	0.4527	0.654	-1.0	0.341	0.5371

Framework versions

• Transformers 4.57.1
• Pytorch 2.8.0+cu128
• Datasets 4.2.0
• Tokenizers 0.22.2