nutrition-extractor

openfoodfacts/nutrition-extractor

nutrition-extractor is a machine learning model from openfoodfacts. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as cc-by-nc-sa-4.0.

This model is a fine-tuned version of microsoft/layoutlmv3-large on the openfoodfacts/nutrient-detection-layout dataset. It allows to automatically extract nutrition values from images of nutrition tables.

Downloads · 30 days

339

7% of all-time downloads

All-time downloads

5.1K

Public

Parameters

357M

5.7 GB on disk

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.onnx1.4 GB · 50%

At a glance

License: cc-by-nc-sa-4.0
Model type: layoutlmv3
Access: Public
Created: Sep 18, 2024
Updated: Jun 16, 2025
SHA: 4c28d479

Base models

microsoft/layoutlmv3-large

Datasets

openfoodfacts/nutrient-detection-layout

Spaces

hiandrewfisher/nutri-extractor-api

Model card

From the Hugging Face model README

nutrition-extractor

It achieves the following results on the evaluation set:

Loss: 0.0534
Precision: 0.9545
Recall: 0.9647
F1: 0.9596
Accuracy: 0.9917

Model description

This model can extract nutrient values from photos of food packaging containing nutrition information. This was developped as part of the Nutrisight project.

For more information about the project, please refer to the nutrisight directory in the openfoodfacts-ai GitHub repository.

As any model using the LayoutLM architecture, this model expects as input:

the image
the tokens (string) on the images
the 2D coordinates of each token

The tokens and their 2D positions is provided by an OCR model. This model was trained using OCR results coming from Google Cloud Vision.

An ONNX export of the model can be found in the onnx directory.

Intended uses & limitations

This model is only intended to be used on images of products where a nutrition table can be found.

Training and evaluation data

The training and evaluation data can be found on the dataset page.

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 1e-05
train_batch_size: 4
eval_batch_size: 4
seed: 42
gradient_accumulation_steps: 8
total_train_batch_size: 32
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
training_steps: 3000
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
1.9852	0.1664	15	1.1500	0.0	0.0	0.0	0.8101
1.0244	0.3329	30	0.8342	0.05	0.0074	0.0129	0.8123
0.7826	0.4993	45	0.6795	0.0789	0.1138	0.0932	0.8479
0.6767	0.6657	60	0.5963	0.1193	0.1644	0.1383	0.8578
0.6031	0.8322	75	0.5406	0.1671	0.2248	0.1917	0.8691
0.5756	0.9986	90	0.4935	0.2291	0.3112	0.2639	0.8778
0.5215	1.1650	105	0.4302	0.3267	0.3948	0.3575	0.8905
0.4782	1.3315	120	0.3782	0.3939	0.4654	0.4267	0.9020
0.4208	1.4979	135	0.3405	0.4027	0.5044	0.4478	0.9081
0.3532	1.6644	150	0.2930	0.4960	0.5820	0.5356	0.9252
0.3458	1.8308	165	0.2658	0.5155	0.6033	0.5560	0.9301
0.302	1.9972	180	0.2321	0.6112	0.7009	0.6530	0.9474
0.2655	2.1637	195	0.2093	0.6471	0.7264	0.6845	0.9520
0.2598	2.3301	210	0.1951	0.7013	0.7557	0.7275	0.9570
0.2364	2.4965	225	0.1794	0.7091	0.7743	0.7402	0.9590
0.2218	2.6630	240	0.1676	0.7216	0.7933	0.7558	0.9621
0.206	2.8294	255	0.1572	0.7436	0.8110	0.7758	0.9650
0.2053	2.9958	270	0.1580	0.7381	0.8114	0.7730	0.9640
0.1876	3.1623	285	0.1406	0.7738	0.8309	0.8013	0.9687
0.1602	3.3287	300	0.1420	0.7714	0.8277	0.7986	0.9671
0.1706	3.4951	315	0.1323	0.7933	0.8379	0.8150	0.9691
0.1585	3.6616	330	0.1313	0.8060	0.8551	0.8298	0.9700
0.1574	3.8280	345	0.1267	0.8129	0.8639	0.8376	0.9717
0.15	3.9945	360	0.1157	0.8336	0.8746	0.8536	0.9754
0.1192	4.1609	375	0.1120	0.8348	0.8709	0.8525	0.9741
0.1313	4.3273	390	0.1130	0.8395	0.8792	0.8589	0.9745
0.1179	4.4938	405	0.1093	0.8370	0.8871	0.8613	0.9755
0.1327	4.6602	420	0.1102	0.8400	0.8853	0.8621	0.9746
0.1323	4.8266	435	0.0997	0.8611	0.8987	0.8795	0.9782
0.1254	4.9931	450	0.0949	0.8499	0.8969	0.8728	0.9775
0.0999	5.1595	465	0.0847	0.8658	0.8992	0.8822	0.9797
0.1017	5.3259	480	0.0803	0.8747	0.9108	0.8924	0.9810
0.091	5.4924	495	0.0796	0.8784	0.9057	0.8918	0.9806
0.0979	5.6588	510	0.0943	0.8607	0.8950	0.8775	0.9773
0.1024	5.8252	525	0.0804	0.8710	0.9062	0.8882	0.9805
0.0952	5.9917	540	0.0787	0.8845	0.9178	0.9008	0.9816
0.0742	6.1581	555	0.0776	0.8918	0.9150	0.9033	0.9823
0.0764	6.3245	570	0.0721	0.9028	0.9187	0.9107	0.9837
0.0813	6.4910	585	0.0664	0.9065	0.9229	0.9146	0.9844
0.0791	6.6574	600	0.0642	0.9026	0.9252	0.9138	0.9848
0.0792	6.8239	615	0.0673	0.8964	0.9248	0.9104	0.9841
0.078	6.9903	630	0.0693	0.8938	0.9224	0.9079	0.9833
0.0678	7.1567	645	0.0672	0.9082	0.9327	0.9203	0.9852
0.0685	7.3232	660	0.0655	0.8926	0.9224	0.9073	0.9840
0.0555	7.4896	675	0.0615	0.9156	0.9271	0.9213	0.9856
0.07	7.6560	690	0.0587	0.9173	0.9373	0.9272	0.9868
0.065	7.8225	705	0.0558	0.9205	0.9405	0.9304	0.9875
0.0599	7.9889	720	0.0579	0.9253	0.9433	0.9342	0.9878
0.0571	8.1553	735	0.0593	0.9148	0.9331	0.9239	0.9866
0.0563	8.3218	750	0.0605	0.9152	0.9322	0.9236	0.9863
0.0602	8.4882	765	0.0581	0.9252	0.9308	0.9280	0.9863
0.0582	8.6546	780	0.0581	0.9206	0.9373	0.9289	0.9872
0.0514	8.8211	795	0.0557	0.9245	0.9382	0.9313	0.9873
0.0467	8.9875	810	0.0520	0.9291	0.9498	0.9394	0.9883
0.0435	9.1540	825	0.0526	0.9229	0.9447	0.9337	0.9880
0.0531	9.3204	840	0.0502	0.9249	0.9443	0.9345	0.9884
0.0502	9.4868	855	0.0545	0.9171	0.9452	0.9309	0.9874
0.0377	9.6533	870	0.0618	0.9077	0.9368	0.9221	0.9851
0.0416	9.8197	885	0.0549	0.9267	0.9392	0.9329	0.9881
0.044	9.9861	900	0.0529	0.9366	0.9475	0.9420	0.9884
0.0383	10.1526	915	0.0490	0.9332	0.9475	0.9403	0.9889
0.0454	10.3190	930	0.0507	0.9264	0.9471	0.9366	0.9885
0.0416	10.4854	945	0.0467	0.9364	0.9498	0.9430	0.9891
0.0403	10.6519	960	0.0499	0.9314	0.9457	0.9385	0.9886
0.0354	10.8183	975	0.0523	0.9258	0.9452	0.9354	0.9883
0.0338	10.9847	990	0.0521	0.9214	0.9424	0.9318	0.9880
0.0347	11.1512	1005	0.0539	0.9235	0.9475	0.9354	0.9880
0.0364	11.3176	1020	0.0560	0.9194	0.9480	0.9335	0.9871
0.0363	11.4840	1035	0.0509	0.9286	0.9480	0.9382	0.9889
0.0308	11.6505	1050	0.0498	0.9389	0.9484	0.9436	0.9893
0.032	11.8169	1065	0.0491	0.9364	0.9443	0.9403	0.9891
0.0331	11.9834	1080	0.0455	0.9373	0.9443	0.9408	0.9892
0.0301	12.1498	1095	0.0486	0.9359	0.9489	0.9423	0.9892
0.0308	12.3162	1110	0.0513	0.9325	0.9503	0.9413	0.9891
0.0253	12.4827	1125	0.0510	0.9296	0.9503	0.9398	0.9892
0.0301	12.6491	1140	0.0533	0.9308	0.9489	0.9397	0.9886
0.0328	12.8155	1155	0.0549	0.9287	0.9443	0.9364	0.9885
0.0298	12.9820	1170	0.0504	0.9402	0.9498	0.9450	0.9895
0.0256	13.1484	1185	0.0515	0.9354	0.9419	0.9387	0.9888
0.0313	13.3148	1200	0.0483	0.9418	0.9545	0.9481	0.9905
0.022	13.4813	1215	0.0463	0.9361	0.9531	0.9445	0.9899
0.0245	13.6477	1230	0.0494	0.9368	0.9494	0.9430	0.9893
0.0251	13.8141	1245	0.0493	0.9404	0.9531	0.9467	0.9898
0.0259	13.9806	1260	0.0511	0.9386	0.9522	0.9454	0.9895
0.03	14.1470	1275	0.0535	0.9344	0.9457	0.9400	0.9889
0.0192	14.3135	1290	0.0491	0.9428	0.9494	0.9461	0.9899
0.0267	14.4799	1305	0.0490	0.9457	0.9545	0.9501	0.9901
0.0241	14.6463	1320	0.0506	0.9435	0.9540	0.9487	0.9899
0.0211	14.8128	1335	0.0510	0.9444	0.9540	0.9492	0.9903
0.0171	14.9792	1350	0.0499	0.9405	0.9545	0.9474	0.9898
0.0226	15.1456	1365	0.0511	0.9366	0.9540	0.9452	0.9894
0.024	15.3121	1380	0.0484	0.9445	0.9559	0.9501	0.9899
0.018	15.4785	1395	0.0482	0.9469	0.9517	0.9493	0.9903
0.0191	15.6449	1410	0.0491	0.9442	0.9512	0.9477	0.9899
0.0203	15.8114	1425	0.0451	0.9510	0.9554	0.9532	0.9912
0.0198	15.9778	1440	0.0447	0.9497	0.9549	0.9523	0.9911
0.0167	16.1442	1455	0.0444	0.9487	0.9540	0.9514	0.9909
0.0178	16.3107	1470	0.0513	0.9386	0.9512	0.9449	0.9892
0.024	16.4771	1485	0.0502	0.9430	0.9536	0.9483	0.9899
0.0206	16.6436	1500	0.0459	0.9483	0.9545	0.9514	0.9908
0.0188	16.8100	1515	0.0469	0.9474	0.9540	0.9507	0.9906
0.016	16.9764	1530	0.0463	0.9468	0.9582	0.9524	0.9906
0.0161	17.1429	1545	0.0455	0.9516	0.9596	0.9556	0.9911
0.0135	17.3093	1560	0.0475	0.9524	0.9573	0.9548	0.9909
0.0148	17.4757	1575	0.0479	0.9440	0.9545	0.9492	0.9905
0.0173	17.6422	1590	0.0455	0.9539	0.9605	0.9572	0.9915
0.0173	17.8086	1605	0.0456	0.9475	0.9554	0.9514	0.9913
0.0185	17.9750	1620	0.0461	0.9498	0.9577	0.9537	0.9908
0.0153	18.1415	1635	0.0472	0.9491	0.9605	0.9548	0.9911
0.0148	18.3079	1650	0.0446	0.9507	0.9587	0.9547	0.9913
0.0136	18.4743	1665	0.0441	0.9486	0.9601	0.9543	0.9914
0.0185	18.6408	1680	0.0478	0.9528	0.9573	0.9551	0.9915
0.0147	18.8072	1695	0.0493	0.9515	0.9652	0.9583	0.9912
0.0156	18.9736	1710	0.0509	0.9440	0.9545	0.9492	0.9903
0.0113	19.1401	1725	0.0460	0.9559	0.9573	0.9566	0.9911
0.014	19.3065	1740	0.0493	0.9439	0.9526	0.9482	0.9905
0.0147	19.4730	1755	0.0498	0.9476	0.9568	0.9522	0.9906
0.0126	19.6394	1770	0.0493	0.9474	0.9531	0.9502	0.9906
0.0167	19.8058	1785	0.0491	0.9463	0.9577	0.9520	0.9904
0.0126	19.9723	1800	0.0474	0.9492	0.9540	0.9516	0.9908
0.0107	20.1387	1815	0.0462	0.9524	0.9577	0.9551	0.9914
0.0115	20.3051	1830	0.0481	0.9504	0.9614	0.9559	0.9911
0.0128	20.4716	1845	0.0486	0.9475	0.9563	0.9519	0.9907
0.0113	20.6380	1860	0.0491	0.9477	0.9591	0.9534	0.9910
0.0119	20.8044	1875	0.0514	0.9494	0.9503	0.9499	0.9901
0.0122	20.9709	1890	0.0480	0.9481	0.9591	0.9536	0.9911
0.0123	21.1373	1905	0.0477	0.9467	0.9577	0.9522	0.9909
0.0116	21.3037	1920	0.0486	0.9485	0.9582	0.9533	0.9910
0.0108	21.4702	1935	0.0488	0.9442	0.9582	0.9511	0.9905
0.0115	21.6366	1950	0.0472	0.9498	0.9587	0.9542	0.9913
0.0083	21.8031	1965	0.0476	0.9490	0.9596	0.9543	0.9911
0.0094	21.9695	1980	0.0475	0.9482	0.9605	0.9543	0.9909
0.0118	22.1359	1995	0.0492	0.9449	0.9554	0.9501	0.9904
0.01	22.3024	2010	0.0486	0.9492	0.9554	0.9523	0.9909
0.0114	22.4688	2025	0.0497	0.9502	0.9577	0.9540	0.9910
0.0091	22.6352	2040	0.0499	0.9503	0.9582	0.9542	0.9910
0.0077	22.8017	2055	0.0502	0.9513	0.9614	0.9563	0.9911
0.01	22.9681	2070	0.0513	0.9544	0.9628	0.9586	0.9913
0.0087	23.1345	2085	0.0485	0.9500	0.9610	0.9554	0.9912
0.0073	23.3010	2100	0.0485	0.9557	0.9628	0.9593	0.9917
0.0083	23.4674	2115	0.0485	0.9535	0.9610	0.9572	0.9913
0.0117	23.6338	2130	0.0479	0.9557	0.9624	0.9590	0.9916
0.0095	23.8003	2145	0.0508	0.9498	0.9587	0.9542	0.9911
0.009	23.9667	2160	0.0513	0.9492	0.9628	0.9560	0.9910
0.0077	24.1331	2175	0.0504	0.9553	0.9628	0.9591	0.9915
0.0087	24.2996	2190	0.0500	0.9521	0.9610	0.9565	0.9913
0.0068	24.4660	2205	0.0506	0.9539	0.9610	0.9574	0.9913
0.0094	24.6325	2220	0.0500	0.9507	0.9591	0.9549	0.9913
0.0088	24.7989	2235	0.0486	0.9508	0.9596	0.9552	0.9914
0.0089	24.9653	2250	0.0507	0.9508	0.9610	0.9559	0.9911
0.0063	25.1318	2265	0.0479	0.9561	0.9610	0.9585	0.9917
0.0058	25.2982	2280	0.0506	0.9526	0.9619	0.9572	0.9911
0.0102	25.4646	2295	0.0499	0.9526	0.9624	0.9575	0.9912
0.0079	25.6311	2310	0.0543	0.9469	0.9614	0.9541	0.9905
0.009	25.7975	2325	0.0498	0.9526	0.9619	0.9572	0.9915
0.0068	25.9639	2340	0.0511	0.9509	0.9619	0.9564	0.9911
0.007	26.1304	2355	0.0492	0.9527	0.9633	0.9580	0.9914
0.0086	26.2968	2370	0.0516	0.9500	0.9610	0.9554	0.9913
0.0078	26.4632	2385	0.0503	0.9504	0.9610	0.9557	0.9914
0.0067	26.6297	2400	0.0514	0.9527	0.9628	0.9577	0.9915
0.0059	26.7961	2415	0.0504	0.9549	0.9628	0.9588	0.9919
0.0089	26.9626	2430	0.0520	0.9517	0.9605	0.9561	0.9916
0.0059	27.1290	2445	0.0512	0.9522	0.9624	0.9573	0.9917
0.0073	27.2954	2460	0.0526	0.9530	0.9610	0.9570	0.9916
0.0065	27.4619	2475	0.0530	0.9527	0.9628	0.9577	0.9916
0.0064	27.6283	2490	0.0515	0.9535	0.9610	0.9572	0.9917
0.0072	27.7947	2505	0.0542	0.9482	0.9610	0.9546	0.9907
0.0066	27.9612	2520	0.0537	0.9491	0.9610	0.9550	0.9909
0.006	28.1276	2535	0.0518	0.9531	0.9628	0.9579	0.9915
0.0074	28.2940	2550	0.0523	0.9521	0.9610	0.9565	0.9914
0.0068	28.4605	2565	0.0534	0.9495	0.9614	0.9555	0.9913
0.0055	28.6269	2580	0.0521	0.9548	0.9619	0.9584	0.9917
0.0056	28.7933	2595	0.0526	0.9522	0.9614	0.9568	0.9913
0.0066	28.9598	2610	0.0527	0.9522	0.9619	0.9570	0.9913
0.0053	29.1262	2625	0.0533	0.9531	0.9628	0.9579	0.9913
0.0063	29.2926	2640	0.0520	0.9530	0.9610	0.9570	0.9913
0.0059	29.4591	2655	0.0533	0.9504	0.9605	0.9554	0.9910
0.0059	29.6255	2670	0.0532	0.9526	0.9619	0.9572	0.9912
0.0062	29.7920	2685	0.0516	0.9535	0.9624	0.9579	0.9917
0.0064	29.9584	2700	0.0515	0.9522	0.9624	0.9573	0.9915
0.0055	30.1248	2715	0.0513	0.9549	0.9633	0.9591	0.9917
0.0064	30.2913	2730	0.0524	0.9540	0.9628	0.9584	0.9916
0.0055	30.4577	2745	0.0530	0.9531	0.9633	0.9582	0.9915
0.0065	30.6241	2760	0.0528	0.9536	0.9642	0.9589	0.9917
0.0068	30.7906	2775	0.0530	0.9518	0.9633	0.9575	0.9916
0.0047	30.9570	2790	0.0545	0.9532	0.9647	0.9589	0.9916
0.0051	31.1234	2805	0.0534	0.9545	0.9647	0.9596	0.9917
0.0044	31.2899	2820	0.0532	0.9531	0.9633	0.9582	0.9914
0.0068	31.4563	2835	0.0532	0.9527	0.9633	0.9580	0.9913
0.0045	31.6227	2850	0.0531	0.9545	0.9638	0.9591	0.9915
0.0047	31.7892	2865	0.0530	0.9540	0.9633	0.9586	0.9916
0.0075	31.9556	2880	0.0533	0.9549	0.9638	0.9593	0.9916
0.0055	32.1221	2895	0.0525	0.9553	0.9638	0.9595	0.9917
0.006	32.2885	2910	0.0523	0.9553	0.9638	0.9595	0.9917
0.0062	32.4549	2925	0.0525	0.9544	0.9633	0.9589	0.9917
0.0059	32.6214	2940	0.0525	0.9549	0.9638	0.9593	0.9917
0.0058	32.7878	2955	0.0531	0.9549	0.9642	0.9596	0.9917
0.005	32.9542	2970	0.0533	0.9536	0.9633	0.9584	0.9916
0.007	33.1207	2985	0.0533	0.9536	0.9633	0.9584	0.9916
0.0047	33.2871	3000	0.0532	0.9536	0.9633	0.9584	0.9916

Framework versions

Transformers 4.40.2
Pytorch 2.5.1
Datasets 2.19.0
Tokenizers 0.19.1

Type: layoutlmv3
License: cc-by-nc-sa-4.0
Created: Sep 18, 2024
Updated: Jun 16, 2025