rtdetr-tray-cart-paper-strict-20260303-213437

This model is a fine-tuned version of PekingU/rtdetr_r101vd_coco_o365 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 7.7488
• Map: 0.5123
• Map 50: 0.8312
• Map 75: 0.5756
• Map Small: 0.5427
• Map Medium: 0.4995
• Map Large: 0.6368
• Mar 1: 0.0606
• Mar 10: 0.352
• Mar 100: 0.6281
• Mar Small: 0.6286
• Mar Medium: 0.6069
• Mar Large: 0.8005
• Map Tray: 0.4236
• Mar 100 Tray: 0.5353
• Map Cart: 0.601
• Mar 100 Cart: 0.721

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: 0.0001
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• 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: cosine
• lr_scheduler_warmup_steps: 3
• num_epochs: 50
• mixed_precision_training: Native AMP

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 Tray	Mar 100 Tray	Map Cart	Mar 100 Cart
No log	1.0	7	24.4270	0.0027	0.0083	0.0011	0.0013	0.006	0.0019	0.0002	0.013	0.1039	0.1714	0.1015	0.0917	0.0019	0.0416	0.0035	0.1661
No log	2.0	14	24.7677	0.0099	0.0318	0.0045	0.0115	0.0213	0.0073	0.006	0.0213	0.1215	0.2429	0.1033	0.2115	0.0178	0.1302	0.002	0.1129
No log	3.0	21	29.9358	0.0257	0.0601	0.0183	0.0389	0.0472	0.021	0.0014	0.0552	0.2108	0.369	0.2031	0.207	0.0434	0.186	0.008	0.2355
No log	4.0	28	18.1668	0.071	0.1455	0.0575	0.1004	0.0849	0.1491	0.0077	0.0838	0.2631	0.419	0.2405	0.3733	0.1304	0.2698	0.0116	0.2565
No log	5.0	35	14.1994	0.1355	0.2703	0.1304	0.3838	0.141	0.1668	0.0314	0.1476	0.383	0.5214	0.3628	0.4838	0.1791	0.3596	0.0918	0.4065
No log	6.0	42	13.0165	0.1973	0.3781	0.1808	0.5304	0.1727	0.386	0.0348	0.1737	0.4277	0.5905	0.3991	0.5893	0.2806	0.4151	0.114	0.4403
No log	7.0	49	10.3521	0.2974	0.566	0.2797	0.5651	0.2726	0.4396	0.0443	0.2392	0.4844	0.65	0.4563	0.6416	0.3526	0.4673	0.2421	0.5016
No log	8.0	56	9.8570	0.3216	0.627	0.3042	0.5535	0.2954	0.4889	0.0467	0.2548	0.4931	0.619	0.4628	0.6819	0.3664	0.4716	0.2768	0.5145
No log	9.0	63	9.2473	0.3408	0.6535	0.3297	0.5577	0.3142	0.5401	0.0443	0.2587	0.5066	0.6381	0.4771	0.6861	0.383	0.4921	0.2987	0.521
No log	10.0	70	9.2426	0.3651	0.6789	0.353	0.5311	0.3415	0.5686	0.0461	0.281	0.5382	0.5976	0.5117	0.7234	0.3606	0.4925	0.3697	0.5839
No log	11.0	77	8.8889	0.3784	0.6815	0.3876	0.5388	0.3518	0.5622	0.0566	0.2766	0.5586	0.6048	0.5411	0.6839	0.3568	0.4931	0.3999	0.6242
No log	12.0	84	8.8623	0.4275	0.7368	0.4325	0.542	0.4082	0.5757	0.0587	0.3091	0.5848	0.6238	0.5612	0.7588	0.3841	0.4953	0.471	0.6742
No log	13.0	91	8.6457	0.4439	0.75	0.4639	0.5296	0.4249	0.5931	0.0543	0.3301	0.5888	0.6333	0.563	0.7776	0.3978	0.5116	0.4901	0.6661
No log	14.0	98	8.6612	0.4525	0.739	0.5128	0.5498	0.4392	0.5629	0.0546	0.3287	0.6136	0.681	0.5929	0.7557	0.3595	0.5239	0.5456	0.7032
No log	15.0	105	8.4938	0.4608	0.7716	0.5252	0.5199	0.4482	0.5838	0.0548	0.3415	0.6005	0.5976	0.5781	0.784	0.3778	0.5091	0.5439	0.6919
No log	16.0	112	8.3763	0.4418	0.7365	0.5038	0.5426	0.4322	0.573	0.057	0.3187	0.5927	0.6143	0.574	0.7381	0.3706	0.508	0.513	0.6774
No log	17.0	119	8.3364	0.4657	0.7701	0.5345	0.5525	0.4544	0.6064	0.0592	0.3308	0.6154	0.6048	0.5916	0.8126	0.3766	0.513	0.5548	0.7177
No log	18.0	126	8.1705	0.4663	0.7863	0.5371	0.5667	0.4535	0.6102	0.0571	0.3357	0.6211	0.619	0.5965	0.82	0.3844	0.5083	0.5482	0.7339
No log	19.0	133	8.2220	0.4546	0.7768	0.4899	0.5599	0.4382	0.6214	0.0556	0.3399	0.6119	0.5952	0.5878	0.8137	0.378	0.5093	0.5312	0.7145
No log	20.0	140	8.1171	0.4784	0.7886	0.5462	0.5823	0.4604	0.6401	0.0572	0.3449	0.6196	0.6262	0.5979	0.7937	0.3769	0.5183	0.5798	0.721
No log	21.0	147	8.0277	0.4701	0.7813	0.5438	0.5729	0.4563	0.6339	0.0564	0.3424	0.6174	0.6167	0.5964	0.7873	0.3875	0.5202	0.5527	0.7145
No log	22.0	154	7.9652	0.4803	0.7996	0.5529	0.5348	0.4674	0.6099	0.0586	0.3424	0.6139	0.6048	0.5945	0.7764	0.3904	0.523	0.5701	0.7048
No log	23.0	161	8.0631	0.4822	0.8089	0.562	0.5207	0.4694	0.5962	0.0531	0.3444	0.6078	0.5881	0.5869	0.784	0.3878	0.5188	0.5767	0.6968
No log	24.0	168	7.9490	0.4739	0.7841	0.5487	0.5397	0.4637	0.6113	0.0543	0.3427	0.625	0.6357	0.604	0.7911	0.3868	0.5226	0.561	0.7274
No log	25.0	175	7.9127	0.4929	0.8043	0.5536	0.572	0.4826	0.611	0.0551	0.3483	0.6328	0.6286	0.6131	0.7972	0.4037	0.527	0.5822	0.7387
No log	26.0	182	7.8650	0.5027	0.8151	0.5898	0.5361	0.4904	0.6398	0.0542	0.3505	0.6304	0.6071	0.6097	0.8092	0.4102	0.5286	0.5952	0.7323
No log	27.0	189	7.8510	0.5112	0.8266	0.5859	0.5593	0.4957	0.6725	0.0644	0.3574	0.627	0.6238	0.6056	0.8031	0.4152	0.5218	0.6072	0.7323
No log	28.0	196	7.8427	0.5127	0.8328	0.575	0.5613	0.4995	0.6797	0.0654	0.3494	0.6232	0.6262	0.6019	0.7952	0.4218	0.5271	0.6035	0.7194
No log	29.0	203	7.9177	0.498	0.8326	0.5616	0.5249	0.4813	0.6596	0.0602	0.3455	0.6102	0.6071	0.5865	0.8013	0.4098	0.5188	0.5862	0.7016
No log	30.0	210	7.8359	0.4855	0.8076	0.5454	0.5357	0.4705	0.6136	0.0608	0.3465	0.6192	0.6214	0.5961	0.8047	0.407	0.5255	0.564	0.7129
No log	31.0	217	7.8632	0.4949	0.8205	0.5677	0.5348	0.4818	0.6333	0.0595	0.3435	0.6245	0.6214	0.6014	0.8118	0.4076	0.5313	0.5822	0.7177
No log	32.0	224	7.8101	0.5021	0.8255	0.5956	0.5453	0.4885	0.6335	0.0608	0.3477	0.6295	0.6262	0.6085	0.8016	0.4052	0.5316	0.5991	0.7274
No log	33.0	231	7.8108	0.4881	0.8102	0.5664	0.5285	0.4709	0.6988	0.06	0.3499	0.6268	0.6167	0.6043	0.8118	0.3946	0.5278	0.5817	0.7258
No log	34.0	238	7.7570	0.4896	0.8106	0.5771	0.5609	0.4781	0.5904	0.0634	0.3366	0.6186	0.6262	0.5947	0.8069	0.3995	0.5307	0.5797	0.7065
No log	35.0	245	7.7686	0.5069	0.8261	0.6068	0.5533	0.4932	0.6384	0.0552	0.3475	0.6263	0.6357	0.6036	0.8069	0.4152	0.53	0.5987	0.7226
No log	36.0	252	7.7488	0.5123	0.8312	0.5756	0.5427	0.4995	0.6368	0.0606	0.352	0.6281	0.6286	0.6069	0.8005	0.4236	0.5353	0.601	0.721

Framework versions

• Transformers 5.2.0
• Pytorch 2.10.0+cu128
• Datasets 4.6.1
• Tokenizers 0.22.2