Mit-b0_corm Open-source Image Segmentation Model - Excellent Evaluation Performance Facilitates Precise Image Segmentation

Mit B0 Corm

Developed by AI-Lab-Makerere

An image segmentation model fine-tuned based on nvidia/mit-b0, demonstrating outstanding performance on the evaluation set with a mean IoU of 0.9210 and mean accuracy of 0.9571.

Image Segmentation

Transformers

Open Source License:Other #High-precision image segmentation #Agricultural damage detection #Multi-class semantic segmentation

Downloads 50

Release Time : 3/5/2025

Model Overview

This model is an image segmentation model fine-tuned on the nvidia/mit-b0 architecture, specifically designed to identify and segment background, bulbs, and damaged areas in images.

Model Features

High-precision segmentation

Achieves a mean IoU of 0.9210 and mean accuracy of 0.9571 on the evaluation set, demonstrating excellent performance.

Multi-class recognition

Accurately identifies and segments three types of targets: background, bulbs, and damaged areas.

Stable training

After 40 training epochs, all metrics stabilized, with validation loss reduced to 0.0433.

Model Capabilities

Image segmentation

Multi-class object recognition

Background region detection

Bulb region detection

Damage region detection

Use Cases

Agricultural inspection

Crop disease detection

Identifies damaged areas in crop images to assist in disease diagnosis.

Damage region detection accuracy reached 93.77%, with an IoU of 89.23%

Botanical research

Plant organ analysis

Precisely segments plant bulb regions for growth condition analysis.

Bulb region detection accuracy reached 93.60%, with an IoU of 87.62%

🚀 mit-b0_corm

This model is a fine - tuned version of nvidia/mit-b0, designed for image segmentation tasks, achieving high accuracy on the evaluation set.

🚀 Quick Start

This model is a fine - tuned version of nvidia/mit-b0 on the None dataset. It achieves the following results on the evaluation set:

Loss: 0.0433
Mean Iou: 0.9210
Mean Accuracy: 0.9571
Overall Accuracy: 0.9853
Accuracy Background: 0.9977
Accuracy Corm: 0.9360
Accuracy Damage: 0.9377
Iou Background: 0.9944
Iou Corm: 0.8762
Iou Damage: 0.8923

🔧 Technical Details

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 6e - 05
train_batch_size: 8
eval_batch_size: 8
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.05
num_epochs: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Background	Accuracy Corm	Accuracy Damage	Iou Background	Iou Corm	Iou Damage
0.933	0.6061	20	1.0299	0.3591	0.6054	0.6910	0.7236	0.1098	0.9827	0.7236	0.0867	0.2671
0.6505	1.2121	40	0.6909	0.6522	0.8240	0.9013	0.9328	0.5651	0.9740	0.9328	0.4509	0.5728
0.4133	1.8182	60	0.4184	0.7567	0.8872	0.9394	0.9609	0.7307	0.9701	0.9607	0.6218	0.6875
0.3299	2.4242	80	0.3451	0.8351	0.9306	0.9617	0.9751	0.8924	0.9243	0.9748	0.7569	0.7735
0.2594	3.0303	100	0.2506	0.8703	0.9412	0.9727	0.9862	0.8989	0.9384	0.9852	0.8019	0.8237
0.2253	3.6364	120	0.2006	0.8851	0.9403	0.9779	0.9939	0.8672	0.9599	0.9915	0.8207	0.8430
0.2222	4.2424	140	0.1654	0.8990	0.9490	0.9805	0.9946	0.9446	0.9079	0.9920	0.8438	0.8612
0.1347	4.8485	160	0.1413	0.9048	0.9508	0.9819	0.9956	0.9334	0.9234	0.9928	0.8526	0.8689
0.1366	5.4545	180	0.1155	0.9094	0.9516	0.9829	0.9966	0.9258	0.9325	0.9933	0.8583	0.8765
0.1121	6.0606	200	0.1086	0.8938	0.9447	0.9801	0.9961	0.9628	0.8753	0.9933	0.8392	0.8487
0.0982	6.6667	220	0.0963	0.9115	0.9524	0.9835	0.9972	0.9374	0.9227	0.9938	0.8626	0.8780
0.0993	7.2727	240	0.0892	0.9094	0.9513	0.9832	0.9968	0.9001	0.9571	0.9940	0.8570	0.8773
0.0813	7.8788	260	0.0842	0.9127	0.9543	0.9837	0.9966	0.9380	0.9281	0.9939	0.8643	0.8798
0.1059	8.4848	280	0.0774	0.9152	0.9541	0.9842	0.9973	0.9258	0.9391	0.9940	0.8673	0.8843
0.082	9.0909	300	0.0729	0.9159	0.9541	0.9843	0.9975	0.9294	0.9355	0.9940	0.8681	0.8854
0.0725	9.6970	320	0.0692	0.9162	0.9544	0.9844	0.9975	0.9247	0.9411	0.9941	0.8686	0.8861
0.0814	10.3030	340	0.0687	0.9161	0.9541	0.9844	0.9975	0.9155	0.9492	0.9942	0.8675	0.8865
0.076	10.9091	360	0.0640	0.9157	0.9555	0.9843	0.9968	0.9219	0.9479	0.9941	0.8680	0.8849
0.07	11.5152	380	0.0633	0.9166	0.9553	0.9845	0.9973	0.9375	0.9310	0.9941	0.8698	0.8859
0.0674	12.1212	400	0.0611	0.9176	0.9549	0.9847	0.9977	0.9217	0.9453	0.9943	0.8704	0.8881
0.0638	12.7273	420	0.0601	0.9116	0.9522	0.9836	0.9977	0.9529	0.9059	0.9941	0.8641	0.8768
0.0566	13.3333	440	0.0582	0.9176	0.9561	0.9847	0.9972	0.9322	0.9387	0.9943	0.8714	0.8872
0.0582	13.9394	460	0.0614	0.9077	0.9502	0.9829	0.9976	0.9583	0.8948	0.9941	0.8588	0.8700
0.0555	14.5455	480	0.0561	0.9146	0.9534	0.9841	0.9978	0.9481	0.9142	0.9941	0.8679	0.8817
0.053	15.1515	500	0.0540	0.9182	0.9551	0.9848	0.9977	0.9185	0.9492	0.9943	0.8707	0.8895
0.059	15.7576	520	0.0549	0.9180	0.9565	0.9848	0.9970	0.9248	0.9478	0.9943	0.8711	0.8887
0.0484	16.3636	540	0.0529	0.9177	0.9563	0.9847	0.9973	0.9405	0.9311	0.9943	0.8721	0.8866
0.0559	16.9697	560	0.0510	0.9192	0.9565	0.9850	0.9974	0.9268	0.9453	0.9943	0.8729	0.8904
0.0542	17.5758	580	0.0512	0.9190	0.9569	0.9850	0.9973	0.9351	0.9382	0.9944	0.8733	0.8894
0.0451	18.1818	600	0.0505	0.9184	0.9557	0.9848	0.9977	0.9428	0.9265	0.9943	0.8729	0.8880
0.05	18.7879	620	0.0499	0.9178	0.9542	0.9848	0.9979	0.9098	0.9549	0.9943	0.8691	0.8899
0.063	19.3939	640	0.0491	0.9190	0.9560	0.9850	0.9975	0.9221	0.9483	0.9943	0.8723	0.8904
0.0484	20.0	660	0.0501	0.9185	0.9569	0.9849	0.9972	0.9427	0.9308	0.9944	0.8732	0.8880
0.0527	20.6061	680	0.0492	0.9186	0.9561	0.9849	0.9976	0.9430	0.9276	0.9943	0.8732	0.8884
0.0583	21.2121	700	0.0476	0.9195	0.9563	0.9851	0.9976	0.9208	0.9506	0.9944	0.8730	0.8911
0.0557	21.8182	720	0.0488	0.9188	0.9565	0.9850	0.9973	0.9191	0.9531	0.9945	0.8723	0.8896
0.0458	22.4242	740	0.0481	0.9194	0.9568	0.9851	0.9973	0.9242	0.9489	0.9944	0.8729	0.8909
0.042	23.0303	760	0.0472	0.9202	0.9570	0.9852	0.9975	0.9326	0.9409	0.9944	0.8749	0.8911
0.0459	23.6364	780	0.0468	0.9191	0.9565	0.9850	0.9976	0.9423	0.9295	0.9944	0.8740	0.8889
0.0491	24.2424	800	0.0464	0.9204	0.9568	0.9852	0.9977	0.9361	0.9366	0.9944	0.8753	0.8914
0.0548	24.8485	820	0.0454	0.9201	0.9565	0.9852	0.9976	0.9244	0.9475	0.9944	0.8740	0.8917
0.0447	25.4545	840	0.0473	0.9176	0.9558	0.9847	0.9976	0.9477	0.9222	0.9944	0.8723	0.8863
0.0457	26.0606	860	0.0468	0.9203	0.9567	0.9852	0.9976	0.9270	0.9456	0.9944	0.8745	0.8922
0.0468	26.6667	880	0.0454	0.9201	0.9572	0.9852	0.9974	0.9403	0.9341	0.9944	0.8753	0.8905
0.0433	27.2727	900	0.0452	0.9208	0.9563	0.9853	0.9980	0.9339	0.9371	0.9943	0.8759	0.8923
0.0438	27.8788	920	0.0452	0.9208	0.9574	0.9853	0.9975	0.9352	0.9396	0.9944	0.8760	0.8920
0.0446	28.4848	940	0.0447	0.9210	0.9568	0.9853	0.9978	0.9349	0.9377	0.9943	0.8760	0.8926
0.0492	29.0909	960	0.0452	0.9211	0.9568	0.9853	0.9978	0.9352	0.9374	0.9943	0.8762	0.8928
0.0481	29.6970	980	0.0456	0.9195	0.9567	0.9851	0.9976	0.9443	0.9283	0.9944	0.8747	0.8893
0.0405	30.3030	1000	0.0447	0.9206	0.9574	0.9853	0.9975	0.9391	0.9355	0.9944	0.8758	0.8916
0.0505	30.9091	1020	0.0443	0.9210	0.9570	0.9853	0.9978	0.9370	0.9364	0.9944	0.8763	0.8923
0.047	31.5152	1040	0.0450	0.9204	0.9568	0.9853	0.9976	0.9223	0.9505	0.9945	0.8744	0.8923
0.0548	32.1212	1060	0.0452	0.9192	0.9561	0.9850	0.9978	0.9442	0.9261	0.9944	0.8744	0.8889
0.0445	32.7273	1080	0.0442	0.9208	0.9573	0.9853	0.9975	0.9320	0.9426	0.9944	0.8758	0.8921
0.0539	33.3333	1100	0.0435	0.9208	0.9571	0.9853	0.9976	0.9359	0.9379	0.9944	0.8758	0.8921
0.0383	33.9394	1120	0.0459	0.9171	0.9549	0.9846	0.9979	0.9493	0.9175	0.9943	0.8716	0.8853
0.0478	34.5455	1140	0.0443	0.9203	0.9572	0.9852	0.9974	0.9246	0.9496	0.9945	0.8748	0.8916
0.0432	35.1515	1160	0.0442	0.9210	0.9571	0.9853	0.9977	0.9349	0.9388	0.9944	0.8762	0.8924
0.0468	35.7576	1180	0.0439	0.9208	0.9572	0.9853	0.9976	0.9371	0.9368	0.9944	0.8761	0.8919
0.0475	36.3636	1200	0.0443	0.9209	0.9571	0.9853	0.9977	0.9371	0.9364	0.9944	0.8762	0.8921
0.0388	36.9697	1220	0.0436	0.9208	0.9573	0.9853	0.9976	0.9371	0.9373	0.9944	0.8761	0.8919
0.0468	37.5758	1240	0.0431	0.9208	0.9574	0.9853	0.9975	0.9343	0.9405	0.9944	0.8760	0.8921
0.0426	38.1818	1260	0.0445	0.9205	0.9570	0.9852	0.9977	0.9415	0.9318	0.9944	0.8758	0.8912
0.0549	38.7879	1280	0.0436	0.9209	0.9571	0.9853	0.9977	0.9373	0.9362	0.9944	0.8761	0.8921
0.045	39.3939	1300	0.0438	0.9208	0.9573	0.9853	0.9976	0.9381	0.9362	0.9944	0.8760	0.8919
0.0287	40.0	1320	0.0433	0.9210	0.9571	0.9853	0.9977	0.9360	0.9377	0.9944	0.8762	0.8923

Framework versions

Transformers 4.44.1
Pytorch 2.6.0+cpu
Datasets 2.21.0
Tokenizers 0.19.1

📄 License

The license type of this model is other.

Property	Details
Library Name	transformers
Model Type	mit - b0_corm
Base Model	nvidia/mit - b0
Tags	vision, image - segmentation, generated_from_trainer

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