Beit-Base-Image-Orientation-Fixer Open-Source Model - Automatically detect and correct image orientation

Beit Base Image Orientation Fixer

Developed by amaye15

Image orientation correction model based on SwinV2 architecture for automatic detection and correction of image orientation

Image Classification

Transformers

Open Source License:Apache-2.0 #Image Orientation Correction #High Precision F1 #SwinV2 Architecture

Downloads 548

Release Time : 9/1/2024

Model Overview

This model is a fine-tuned image orientation corrector based on the SwinV2 architecture, capable of automatically identifying and correcting incorrect image orientations, achieving an F1 score of 0.9391 on the evaluation set.

Model Features

High-precision Orientation Recognition

Achieves an F1 score of 0.9391 on the evaluation set, accurately identifying image orientation

Based on SwinV2 Architecture

Utilizes the advanced Swin Transformer V2 architecture with excellent visual feature extraction capabilities

Efficient Processing Capability

Supports input resolution of 256x256 pixels, efficiently processing medium-sized images

Model Capabilities

Image Orientation Detection

Automatic Image Rotation

Image Orientation Classification

Use Cases

Image Processing

Automatic Correction of User-uploaded Images

Automatically detects and corrects the orientation of user-uploaded images, improving user experience

Reduces orientation errors in images by over 90%

Image Database Organization

Batch processing of incorrectly oriented images in an image database

Improves image retrieval and browsing efficiency

🚀 SwinV2-Base-Image-Orientation-Fixer

This model is a fine - tuned version of [microsoft/swinv2 - base - patch4 - window16 - 256](https://huggingface.co/microsoft/swinv2 - base - patch4 - window16 - 256) on the None dataset. It can be used to fix image orientation, achieving high accuracy on the evaluation set.

🚀 Quick Start

This model is a fine - tuned version of [microsoft/swinv2 - base - patch4 - window16 - 256](https://huggingface.co/microsoft/swinv2 - base - patch4 - window16 - 256) on the None dataset. It achieves the following results on the evaluation set:

Loss: 0.1295
F1: 0.9391

✨ Features

Fine - tuned Model: Based on the pre - trained microsoft/swinv2 - base - patch4 - window16 - 256, it is fine - tuned for image orientation fixing.
High Performance: Achieves a low loss of 0.1295 and a high F1 score of 0.9391 on the evaluation set.

🔧 Technical Details

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 5e - 05
train_batch_size: 32
eval_batch_size: 32
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
lr_scheduler_type: linear
training_steps: 10000

Training results

Training Loss	Epoch	Step	Validation Loss	F1
1.3053	0.001	10	1.1901	0.2711
1.1711	0.002	20	1.1128	0.2697
1.1158	0.003	30	1.0354	0.3505
1.0164	0.004	40	0.9894	0.3356
1.0899	0.005	50	0.9472	0.3613
0.9108	0.006	60	0.9176	0.4340
0.9161	0.007	70	0.9004	0.3447
0.8827	0.008	80	0.8459	0.5296
0.9068	0.009	90	0.7963	0.5104
0.8606	0.01	100	0.7538	0.5552
0.815	0.011	110	0.7165	0.5863
0.758	0.012	120	0.6884	0.5611
0.7998	0.013	130	0.6822	0.6414
0.8375	0.014	140	0.6660	0.5511
0.7375	0.015	150	0.6145	0.6981
0.6883	0.016	160	0.5882	0.6456
0.6961	0.017	170	0.5733	0.6803
0.575	0.018	180	0.5161	0.7218
0.4773	0.019	190	0.6457	0.6837
0.6862	0.02	200	0.5336	0.7211
0.5798	0.021	210	0.4840	0.7517
0.5621	0.022	220	0.4435	0.7674
0.475	0.023	230	0.4663	0.7113
0.6144	0.024	240	0.5058	0.6841
0.5164	0.025	250	0.4790	0.7295
0.524	0.026	260	0.4770	0.6971
0.5406	0.027	270	0.4121	0.7509
0.4378	0.028	280	0.4170	0.7875
0.5211	0.029	290	0.4221	0.7851
0.4258	0.03	300	0.3769	0.7990
0.4177	0.031	310	0.3696	0.8143
0.4183	0.032	320	0.3798	0.8072
0.487	0.033	330	0.4113	0.8098
0.3602	0.034	340	0.3808	0.8059
0.4247	0.035	350	0.3768	0.8271
0.4483	0.036	360	0.3503	0.7964
0.3606	0.037	370	0.3605	0.8314
0.3878	0.038	380	0.3359	0.8390
0.4085	0.039	390	0.2981	0.8606
0.3672	0.04	400	0.3072	0.8511
0.3539	0.041	410	0.3090	0.8481
0.4045	0.042	420	0.3052	0.8555
0.403	0.043	430	0.3610	0.8221
0.3892	0.044	440	0.3189	0.8396
0.4989	0.045	450	0.3337	0.8287
0.3922	0.046	460	0.3019	0.8540
0.3794	0.047	470	0.3157	0.8476
0.4158	0.048	480	0.3050	0.8553
0.3367	0.049	490	0.2884	0.8615
0.3991	0.05	500	0.3451	0.8407
0.411	0.051	510	0.2762	0.8623
0.2855	0.052	520	0.2766	0.8701
0.4612	0.053	530	0.2799	0.8712
0.4118	0.054	540	0.3085	0.8511
0.2906	0.055	550	0.2841	0.8759
0.3242	0.056	560	0.2719	0.8823
0.3575	0.057	570	0.3384	0.8561
0.3562	0.058	580	0.2722	0.8791
0.3944	0.059	590	0.3507	0.8465
0.3175	0.06	600	0.2901	0.8668
0.3608	0.061	610	0.2902	0.8660
0.2619	0.062	620	0.3373	0.8543
0.3243	0.063	630	0.2703	0.8884
0.361	0.064	640	0.2891	0.8662
0.3267	0.065	650	0.2739	0.8784
0.261	0.066	660	0.2602	0.8747
0.2521	0.067	670	0.2641	0.8883
0.391	0.068	680	0.2589	0.8870
0.3604	0.069	690	0.2622	0.8903
0.2983	0.07	700	0.2528	0.8846
0.2521	0.071	710	0.2571	0.8916
0.4368	0.072	720	0.2839	0.8877
0.3208	0.073	730	0.2898	0.8742
0.2887	0.074	740	0.2700	0.8839
0.3075	0.075	750	0.2707	0.8770
0.3465	0.076	760	0.2828	0.8695
0.2863	0.077	770	0.2874	0.8823
0.3402	0.078	780	0.2782	0.8781
0.3495	0.079	790	0.2538	0.8929
0.3177	0.08	800	0.2437	0.8779
0.3012	0.081	810	0.2865	0.8837
0.4079	0.082	820	0.2573	0.8830
0.2915	0.083	830	0.3135	0.8707
0.2407	0.084	840	0.2804	0.8844
0.2574	0.085	850	0.2810	0.8713
0.3141	0.086	860	0.2827	0.8802
0.2601	0.087	870	0.3076	0.8693
0.3462	0.088	880	0.2588	0.8714
0.3356	0.089	890	0.2677	0.8761
0.3135	0.09	900	0.2715	0.8740
0.369	0.091	910	0.2674	0.8705
0.2866	0.092	920	0.2617	0.8827
0.251	0.093	930	0.2483	0.8894
0.1822	0.094	940	0.2679	0.8817
0.2569	0.095	950	0.2810	0.8847
0.3046	0.096	960	0.2774	0.8773
0.2099	0.097	970	0.2738	0.8715
0.2961	0.098	980	0.2603	0.8860
0.2724	0.099	990	0.2661	0.8813
0.3179	0.1	1000	0.2414	0.8837
0.3635	0.101	1010	0.2433	0.8916
0.2815	0.102	1020	0.2562	0.8784
0.2758	0.103	1030	0.2358	0.8941
0.2664	0.104	1040	0.2571	0.8919
0.2584	0.105	1050	0.2617	0.8758
0.3165	0.106	1060	0.2690	0.8834
0.2877	0.107	1070	0.2362	0.8926
0.2713	0.108	1080	0.2416	0.8905
0.2598	0.109	1090	0.2525	0.8806
0.2796	0.11	1100	0.2433	0.8950
0.2558	0.111	1110	0.2562	0.8930
0.2443	0.112	1120	0.2714	0.8838
0.3383	0.113	1130	0.2387	0.8971
0.2636	0.114	1140	0.2721	0.8673
0.2851	0.115	1150	0.2459	0.8830
0.2072	0.116	1160	0.2309	0.8929
0.2331	0.117	1170	0.2942	0.8835
0.2361	0.118	1180	0.2517	0.8958
0.3166	0.119	1190	0.2590	0.8947
0.2891	0.12	1200	0.2725	0.8787
0.3136	0.121	1210	0.2321	0.8969
0.2569	0.122	1220	0.2489	0.8929
0.262	0.123	1230	0.2493	0.8869
0.311	0.124	1240	0.2285	0.8996
0.2848	0.125	1250	0.2407	0.8877
0.2321	0.126	1260	0.2276	0.8887
0.2398	0.127	1270	0.2598	0.8879
0.2399	0.128	1280	0.2331	0.8850
0.3352	0.129	1290	0.2384	0.8908
0.3042	0.13	1300	0.2161	0.9016
0.22	0.131	1310	0.2493	0.8815
0.2821	0.132	1320	0.2159	0.8989
0.2903	0.133	1330	0.2258	0.8990
0.3207	0.134	1340	0.2223	0.9039
0.204	0.135	1350	0.2109	0.9000
0.2346	0.136	1360	0.2305	0.8923
0.2775	0.137	1370	0.2150	0.9013
0.2689	0.138	1380	0.2313	0.9031
0.2346	0.139	1390	0.2181	0.9017
0.2454	0.14	1400	0.2273	0.9002
0.2867	0.141	1410	0.2218	0.8954
0.3079	0.142	1420	0.2302	0.8858
0.2169	0.143	1430	0.2588	0.8806
0.3228	0.144	1440	0.2274	0.8998
0.3602	0.145	1450	0.2293	0.8955
0.2999	0.146	1460	0.2218	0.8977
0.2667	0.147	1470	0.2312	0.8941
0.2569	0.148	1480	0.2269	0.8991
0.1956	0.149	1490	0.2411	0.8928
0.3543	0.15	1500	0.2184	0.8998
0.2969	0.151	1510	0.2328	0.8989
0.297	0.152	1520	0.2190	0.9038
0.258	0.153	1530	0.2292	0.9037
0.2046	0.154	1540	0.2245	0.8946
0.224	0.155	1550	0.2213	0.9004
0.2647	0.156	1560	0.2498	0.8844
0.2191	0.157	1570	0.2382	0.9008
0.2515	0.158	1580	0.2387	0.8952
0.2661	0.159	1590	0.2342	0.8864
0.2301	0.16	1600	0.2692	0.8745
0.2119	0.161	1610	0.2365	0.8897
0.1666	0.162	1620	0.2417	0.8896
0.2507	0.163	1630	0.2416	0.8873
0.2006	0.164	1640	0.2659	0.8839
0.1649	0.165	1650	0.2301	0.8972
0.2099	0.166	1660	0.2514	0.8956
0.3191	0.167	1670	0.2337	0.8846
0.2718	0.168	1680	0.2297	0.9123
0.2827	0.169	1690	0.2338	0.8931
0.2433	0.17	1700	0.2308	0.8927
0.2719	0.171	1710	0.2331	0.8946
0.2151	0.172	1720	0.2131	0.9052
0.2758	0.173	1730	0.2272	0.8989
0.3078	0.174	1740	0.2180	0.9072
0.3012	0.175	1750	0.2258	0.9034
0.3162	0.176	1760	0.2213	0.9057
0.2551	0.177	1770	0.2595	0.8818
0.2563	0.178	1780	0.2230	0.8973
0.2825	0.179	1790	0.2222	0.8957
0.1916	0.18	1800	0.2307	0.8981
0.2145	0.181	1810	0.2285	0.8975
0.2029	0.182	1820	0.2357	0.9003
0.1559	0.183	1830	0.2645	0.8880
0.3173	0.184	1840	0.2209	0.8992
0.2343	0.185	1850	0.2334	0.9043
0.233	0.186	1860	0.2241	0.9025
0.2697	0.187	1870	0.2150	0.8991
0.3023	0.188	1880	0.2312	0.8976
0.2052	0.189	1890	0.2053	0.8975
0.2601	0.19	1900	0.2042	0.8974
0.2966	0.191	1910	0.2142	0.8971
0.2338	0.192	1920

📄 License

This model is licensed under the apache - 2.0 license.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご