detr_finetuned_cppe5

This model is a fine-tuned version of microsoft/conditional-detr-resnet-50 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.8041
• Map: 0.4041
• Map 50: 0.8246
• Map 75: 0.3402
• Map Small: 0.3079
• Map Medium: 0.3562
• Map Large: 0.6364
• Mar 1: 0.1839
• Mar 10: 0.4794
• Mar 100: 0.5657
• Mar Small: 0.4329
• Mar Medium: 0.5174
• Mar Large: 0.7856
• Map Hardhat: 0.4075
• Mar 100 Hardhat: 0.5473
• Map No-hardhat: 0.4007
• Mar 100 No-hardhat: 0.5842

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 with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• num_epochs: 30

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 Hardhat	Mar 100 Hardhat	Map No-hardhat	Mar 100 No-hardhat
No log	1.0	125	1.1847	0.0822	0.1954	0.0553	0.0796	0.0937	0.2324	0.1312	0.3509	0.4354	0.2086	0.3775	0.6788	0.1426	0.5655	0.0218	0.3053
No log	2.0	250	1.0931	0.1205	0.2648	0.0965	0.1277	0.1601	0.1647	0.1587	0.3817	0.4625	0.2886	0.435	0.6636	0.1868	0.5618	0.0542	0.3632
No log	3.0	375	1.1882	0.1449	0.3805	0.0834	0.0775	0.1623	0.3615	0.1042	0.3262	0.4281	0.3086	0.347	0.5962	0.2321	0.5036	0.0577	0.3526
1.472	4.0	500	1.0418	0.2419	0.6022	0.1416	0.1374	0.3265	0.3723	0.1444	0.4138	0.4827	0.2986	0.5035	0.6114	0.2844	0.5127	0.1994	0.4526
1.472	5.0	625	1.0232	0.2349	0.5815	0.1564	0.1736	0.3006	0.3826	0.1395	0.3992	0.4974	0.2686	0.5149	0.6689	0.2891	0.5527	0.1807	0.4421
1.472	6.0	750	0.9985	0.293	0.6561	0.2108	0.2129	0.3184	0.426	0.1604	0.4373	0.5125	0.3129	0.5225	0.6879	0.3452	0.5618	0.2407	0.4632
1.472	7.0	875	0.9616	0.3145	0.7258	0.2615	0.2491	0.3435	0.4999	0.1381	0.4634	0.5455	0.3171	0.5568	0.7386	0.3164	0.5436	0.3126	0.5474
0.9939	8.0	1000	0.9688	0.3194	0.7461	0.1786	0.1922	0.3171	0.576	0.16	0.4325	0.5252	0.2943	0.5194	0.7462	0.2982	0.4873	0.3405	0.5632
0.9939	9.0	1125	0.9211	0.3572	0.7888	0.2944	0.2046	0.3196	0.5454	0.1482	0.4632	0.5346	0.2571	0.5625	0.7303	0.3545	0.5218	0.3599	0.5474
0.9939	10.0	1250	0.9664	0.3463	0.7569	0.2541	0.2503	0.3473	0.4924	0.1611	0.4344	0.5116	0.31	0.499	0.7258	0.3423	0.5127	0.3504	0.5105
0.9939	11.0	1375	0.9463	0.3261	0.8263	0.2012	0.2582	0.2676	0.5931	0.1643	0.4153	0.5134	0.32	0.4437	0.7629	0.3386	0.5164	0.3136	0.5105
0.8775	12.0	1500	0.9153	0.3571	0.7972	0.2882	0.2397	0.3273	0.5803	0.1587	0.4377	0.5556	0.3571	0.566	0.7189	0.3347	0.5164	0.3795	0.5947
0.8775	13.0	1625	0.9063	0.3512	0.8299	0.2471	0.2342	0.3119	0.6117	0.1695	0.4222	0.5016	0.2786	0.4672	0.7439	0.3422	0.4927	0.3602	0.5105
0.8775	14.0	1750	0.9384	0.3351	0.7633	0.2393	0.1938	0.3064	0.5551	0.1723	0.4257	0.5105	0.3371	0.4718	0.7076	0.3496	0.5	0.3206	0.5211
0.8775	15.0	1875	0.8734	0.3836	0.8279	0.3055	0.2541	0.348	0.614	0.1748	0.4373	0.531	0.3729	0.4941	0.7386	0.3671	0.5145	0.4002	0.5474
0.7888	16.0	2000	0.8470	0.3763	0.8437	0.2603	0.2854	0.3289	0.5821	0.1822	0.4556	0.5455	0.4314	0.4861	0.7568	0.3894	0.5436	0.3633	0.5474
0.7888	17.0	2125	0.8579	0.3708	0.8189	0.2792	0.2701	0.2976	0.6115	0.185	0.443	0.5206	0.3957	0.4633	0.7326	0.3986	0.5255	0.343	0.5158
0.7888	18.0	2250	0.8404	0.3714	0.7962	0.2522	0.2531	0.3327	0.6139	0.1778	0.4587	0.5433	0.3729	0.5084	0.7455	0.3709	0.5182	0.3719	0.5684
0.7888	19.0	2375	0.8268	0.3997	0.8285	0.3108	0.2915	0.3526	0.5942	0.1829	0.4882	0.5481	0.4157	0.4986	0.7644	0.4014	0.5436	0.3979	0.5526
0.7048	20.0	2500	0.8091	0.4209	0.8122	0.4316	0.2668	0.377	0.6569	0.1964	0.4669	0.5568	0.4	0.5206	0.7462	0.4154	0.54	0.4265	0.5737
0.7048	21.0	2625	0.8206	0.416	0.8227	0.303	0.3221	0.3747	0.6208	0.1839	0.4811	0.5401	0.3729	0.4992	0.747	0.4198	0.5382	0.4121	0.5421
0.7048	22.0	2750	0.8108	0.4266	0.8502	0.4021	0.3038	0.3847	0.6317	0.1965	0.4688	0.5534	0.4257	0.5125	0.7515	0.4264	0.5436	0.4269	0.5632
0.7048	23.0	2875	0.8239	0.4103	0.8158	0.3492	0.2874	0.3626	0.6316	0.1919	0.4572	0.5533	0.4114	0.5152	0.7462	0.417	0.5382	0.4036	0.5684
0.6439	24.0	3000	0.8092	0.4077	0.825	0.3504	0.3205	0.3525	0.6074	0.1893	0.4883	0.5641	0.4357	0.5228	0.7652	0.4129	0.5545	0.4026	0.5737
0.6439	25.0	3125	0.8076	0.4104	0.8432	0.3547	0.316	0.3559	0.6302	0.1893	0.4689	0.5535	0.4429	0.5027	0.7515	0.4187	0.5491	0.4021	0.5579
0.6439	26.0	3250	0.7988	0.4133	0.837	0.3469	0.3285	0.3631	0.6222	0.2035	0.4849	0.573	0.4643	0.5166	0.7902	0.4219	0.5618	0.4048	0.5842
0.6439	27.0	3375	0.8015	0.4082	0.832	0.3262	0.3104	0.3619	0.62	0.1699	0.4785	0.5693	0.4429	0.5174	0.7902	0.4165	0.5491	0.3998	0.5895
0.6079	28.0	3500	0.8043	0.4064	0.824	0.3412	0.3052	0.3588	0.6301	0.1857	0.4785	0.5631	0.4329	0.5111	0.7856	0.4145	0.5473	0.3982	0.5789
0.6079	29.0	3625	0.8043	0.403	0.8246	0.3378	0.3082	0.3533	0.6358	0.1839	0.4794	0.5667	0.44	0.5174	0.7856	0.4076	0.5491	0.3984	0.5842
0.6079	30.0	3750	0.8041	0.4041	0.8246	0.3402	0.3079	0.3562	0.6364	0.1839	0.4794	0.5657	0.4329	0.5174	0.7856	0.4075	0.5473	0.4007	0.5842

Framework versions

• Transformers 4.51.3
• Pytorch 2.7.0+cu126
• Datasets 3.6.0
• Tokenizers 0.21.1