segformer_mixed-set2-788images_mit-b5_RGB_15Epochs

Hasano20/segformer_mixed-set2-788images_mit-b5_RGB_15Epochs

segformer_mixed-set2-788images_mit-b5_RGB_15Epochs is a image segmentation model from Hasano20. Use it for the image segmentation task on the model card, and read the license before you ship it in a product. It is set up for transformers. The card lists the license as other.

should probably proofread and complete it, then remove this comment. --

Downloads · 30 days

6% of all-time downloads

All-time downloads

Public

Parameters

84.6M

339 MB on disk

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.safetensors339 MB · 100%

At a glance

Task: Image Segmentation
Library: transformers
License: other
Model type: segformer
Access: Public
Created: Jun 25, 2024
Updated: Jun 27, 2024
SHA: 3add9937

Base models

nvidia/mit-b5

Model card

From the Hugging Face model README

SegFormer_Mixed_Set2_788images_mit-b5_RGB

This model is a fine-tuned version of nvidia/mit-b5 on the Hasano20/Mixed_Set2_788images dataset. It achieves the following results on the evaluation set:

Train-Loss: 0.0099
Loss: 0.0150
Mean Iou: 0.9788
Mean Accuracy: 0.9887
Overall Accuracy: 0.9948
Accuracy Background: 0.9958
Accuracy Melt: 0.9735
Accuracy Substrate: 0.9969
Iou Background: 0.9926
Iou Melt: 0.9509
Iou Substrate: 0.9927

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: 0.0001
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: cosine
lr_scheduler_warmup_steps: 100
num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Background	Accuracy Melt	Accuracy Substrate	Iou Background	Iou Melt	Iou Substrate
0.1619	0.7042	50	0.1799	0.7782	0.8306	0.9444	0.9902	0.5371	0.9645	0.9436	0.4720	0.9192
0.062	1.4085	100	0.1065	0.8361	0.8630	0.9638	0.9833	0.6084	0.9972	0.9720	0.5922	0.9441
0.1757	2.1127	150	0.1157	0.8551	0.8896	0.9617	0.9803	0.7065	0.9820	0.9484	0.6731	0.9438
0.0872	2.8169	200	0.0446	0.9302	0.9539	0.9844	0.9938	0.8760	0.9920	0.9846	0.8282	0.9777
0.0336	3.5211	250	0.0338	0.9469	0.9751	0.9877	0.9913	0.9431	0.9910	0.9857	0.8719	0.9831
0.0417	4.2254	300	0.0488	0.9281	0.9820	0.9830	0.9941	0.9765	0.9753	0.9877	0.8233	0.9732
0.0273	4.9296	350	0.0295	0.9516	0.9628	0.9892	0.9952	0.8960	0.9973	0.9895	0.8819	0.9835
0.0249	5.6338	400	0.0228	0.9627	0.9807	0.9913	0.9916	0.9544	0.9960	0.9890	0.9112	0.9879
0.0247	6.3380	450	0.0234	0.9642	0.9886	0.9915	0.9919	0.9814	0.9925	0.9894	0.9151	0.9881
0.0219	7.0423	500	0.0220	0.9656	0.9768	0.9920	0.9943	0.9386	0.9975	0.9908	0.9178	0.9882
0.0172	7.7465	550	0.0206	0.9672	0.9888	0.9923	0.9951	0.9792	0.9919	0.9913	0.9215	0.9888
0.018	8.4507	600	0.0169	0.9747	0.9859	0.9937	0.9944	0.9665	0.9969	0.9910	0.9420	0.9911
0.0152	9.1549	650	0.0180	0.9726	0.9856	0.9932	0.9968	0.9659	0.9942	0.9909	0.9366	0.9902
0.016	9.8592	700	0.0180	0.9729	0.9877	0.9936	0.9955	0.9726	0.9949	0.9917	0.9360	0.9909
0.0132	10.5634	750	0.0169	0.9746	0.9872	0.9938	0.9944	0.9708	0.9965	0.9914	0.9410	0.9913
0.0115	11.2676	800	0.0156	0.9761	0.9898	0.9941	0.9952	0.9789	0.9954	0.9920	0.9446	0.9917
0.0143	11.9718	850	0.0155	0.9765	0.9895	0.9943	0.9962	0.9772	0.9952	0.9923	0.9452	0.9920
0.0106	12.6761	900	0.0146	0.9778	0.9898	0.9946	0.9959	0.9777	0.9959	0.9924	0.9485	0.9925
0.0106	13.3803	950	0.0146	0.9780	0.9888	0.9947	0.9967	0.9736	0.9959	0.9923	0.9490	0.9928
0.0068	14.0845	1000	0.0147	0.9784	0.9883	0.9947	0.9966	0.9718	0.9964	0.9924	0.9501	0.9928
0.0115	14.7887	1050	0.0163	0.9759	0.9901	0.9942	0.9958	0.9795	0.9950	0.9925	0.9436	0.9917
0.0099	15.4930	1100	0.0150	0.9788	0.9887	0.9948	0.9958	0.9735	0.9969	0.9926	0.9509	0.9927

Framework versions

Transformers 4.41.2
Pytorch 2.0.1+cu117
Datasets 2.19.2
Tokenizers 0.19.1

Task: Image Segmentation
Library: transformers
Type: segformer
License: other
Created: Jun 25, 2024
Updated: Jun 27, 2024