detr-resnet-50-license-plates

This model is a fine-tuned version of facebook/detr-resnet-50 on the merve/license-plates dataset. It achieves the following results on the evaluation set:

• Loss: 0.6218
• Map: 0.5566
• Map 50: 0.9043
• Map 75: 0.6949
• Map Small: 0.3178
• Map Medium: 0.6551
• Map Large: 0.7411
• Mar 1: 0.6056
• Mar 10: 0.6215
• Mar 100: 0.6299
• Mar Small: 0.403
• Mar Medium: 0.7282
• Mar Large: 0.8
• Map License Plate: 0.5566
• Mar 100 License Plate: 0.6299

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
• num_epochs: 15.0

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 License Plate	Mar 100 License Plate
No log	1.0	78	1.2107	0.0487	0.1259	0.0233	0.0062	0.074	0.0124	0.1617	0.3776	0.457	0.2606	0.5366	0.7333	0.0487	0.457
No log	2.0	156	0.9759	0.1636	0.3357	0.112	0.0177	0.2458	0.1747	0.4056	0.5178	0.5196	0.2515	0.6366	0.7	0.1636	0.5196
No log	3.0	234	0.8988	0.3416	0.6219	0.3061	0.0322	0.4964	0.4529	0.4393	0.5271	0.5299	0.2667	0.6437	0.7333	0.3416	0.5299
No log	4.0	312	1.0295	0.326	0.7418	0.2184	0.055	0.4436	0.4	0.3813	0.4206	0.4336	0.1818	0.5451	0.5667	0.326	0.4336
No log	5.0	390	0.8468	0.4294	0.7961	0.4045	0.1182	0.5614	0.5099	0.4888	0.5196	0.5196	0.2152	0.6507	0.7667	0.4294	0.5196
No log	6.0	468	0.8805	0.483	0.8334	0.5289	0.1858	0.6005	0.6822	0.5308	0.5729	0.5832	0.3333	0.6887	0.8333	0.483	0.5832
0.9821	7.0	546	0.7494	0.5054	0.8775	0.5352	0.2229	0.6195	0.6294	0.5533	0.5776	0.5804	0.3182	0.693	0.8	0.5054	0.5804
0.9821	8.0	624	0.7213	0.4956	0.8559	0.5254	0.2157	0.6127	0.6629	0.5542	0.5794	0.5832	0.303	0.7028	0.8333	0.4956	0.5832
0.9821	9.0	702	0.7030	0.529	0.881	0.6358	0.2547	0.6426	0.7243	0.585	0.5991	0.6019	0.3394	0.7155	0.8	0.529	0.6019
0.9821	10.0	780	0.7783	0.48	0.8672	0.4323	0.2127	0.5944	0.6192	0.5421	0.5607	0.5617	0.2939	0.6789	0.7333	0.48	0.5617
0.9821	11.0	858	0.7211	0.5355	0.8775	0.6051	0.2605	0.6533	0.666	0.5841	0.6019	0.6028	0.3273	0.7225	0.8	0.5355	0.6028
0.9821	12.0	936	0.6756	0.527	0.8723	0.6	0.2579	0.6304	0.7384	0.5813	0.6056	0.6084	0.3727	0.7085	0.8333	0.527	0.6084
0.6598	13.0	1014	0.6643	0.5319	0.9077	0.5504	0.2811	0.6338	0.7243	0.586	0.6047	0.6121	0.3788	0.7127	0.8	0.5319	0.6121
0.6598	14.0	1092	0.6265	0.5527	0.9038	0.6582	0.3069	0.6504	0.7243	0.5944	0.6187	0.6224	0.403	0.7169	0.8	0.5527	0.6224
0.6598	15.0	1170	0.6218	0.5566	0.9043	0.6949	0.3178	0.6551	0.7411	0.6056	0.6215	0.6299	0.403	0.7282	0.8	0.5566	0.6299

Framework versions

• Transformers 5.3.0.dev0
• Pytorch 2.10.0+cu128
• Datasets 4.5.0
• Tokenizers 0.22.2