{"id":13738550,"url":"https://github.com/prakashjayy/pytorch_classifiers","last_synced_at":"2025-05-08T16:34:34.139Z","repository":{"id":75299917,"uuid":"112632626","full_name":"prakashjayy/pytorch_classifiers","owner":"prakashjayy","description":"Almost any Image classification problem using pytorch","archived":false,"fork":false,"pushed_at":"2020-10-01T23:53:20.000Z","size":59,"stargazers_count":132,"open_issues_count":7,"forks_count":41,"subscribers_count":10,"default_branch":"master","last_synced_at":"2024-11-15T07:34:48.069Z","etag":null,"topics":["deep-learning","image-classification","machine-learning","python","pytorch","transfer-learning"],"latest_commit_sha":null,"homepage":"https://medium.com/@14prakash/almost-any-image-classification-problem-using-pytorch-i-am-in-love-with-pytorch-26c7aa979ec4","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"gpl-3.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/prakashjayy.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null}},"created_at":"2017-11-30T16:15:17.000Z","updated_at":"2024-11-12T08:46:24.000Z","dependencies_parsed_at":"2023-03-29T05:18:21.034Z","dependency_job_id":null,"html_url":"https://github.com/prakashjayy/pytorch_classifiers","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/prakashjayy%2Fpytorch_classifiers","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/prakashjayy%2Fpytorch_classifiers/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/prakashjayy%2Fpytorch_classifiers/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/prakashjayy%2Fpytorch_classifiers/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/prakashjayy","download_url":"https://codeload.github.com/prakashjayy/pytorch_classifiers/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":253105659,"owners_count":21855071,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["deep-learning","image-classification","machine-learning","python","pytorch","transfer-learning"],"created_at":"2024-08-03T03:02:26.498Z","updated_at":"2025-05-08T16:34:33.857Z","avatar_url":"https://github.com/prakashjayy.png","language":"Python","funding_links":[],"categories":["Python"],"sub_categories":[],"readme":"## This is an experimental setup to build code base for pytorch. Its main is to experiment faster using transfer learning on all available pre-trained models.\n\n### Dataset: Plant Seedlings classification\n\nClasses present:\n-----------------\n- Black-grass\n- Charlock\n- Cleavers\n- Common Chickweed\n- Common wheat\n- Fat Hen\n- Loose Silky-bent\n- Maize\n- Scentless Mayweed\n- Shepherds Purse\n- Small-flowered Cranesbill\n- Sugar beet\n\n#### Run the following commands\nTo train the models:\n```\nmake -f Makefile.train\n```\n\nTo test the models:\n```\nmake -f Makefile.predict\n```\n\n\n## Results\ntotal number of train images: 4268\ntotal number of val images: 482\ntotal number of test images: 794\n\n\n# Results with Full-Agumentation strategy:\n\n## Trained the networks in three methods:\n   - Full Finetuning\n   - Freeze first few layers\n\n## Case-1 - Finetuning entire network\nModels           | Train Accuracy_score |    Val Accuracy_score |\n-------------    | :------------------: | --------------------: |\nresnet18         |    0.92783           |    0.93153            |\nresnet34         |    0.9522            |    0.94190            |\nresnet50         |    0.95665           |    0.94398            |\nresnet101        |    0.96696           |    0.96265            |\nresnet152        |    0.96555           |    0.95643            |\nsqueezenet1_0    |    0.94329           |    0.92738            |\nsqueezenet1_1    |    0.93955           |    0.93153            |\ndensenet121      |    0.95243           |    0.9336             |\ndensenet169      |    0.96626           |    0.93983            |\ndensenet201      |    0.96063           |    0.95020            |\ninception_v3     |    0.94212           |    0.93568            |\nvgg11            |    0.93814           |    0.93153            |\nvgg13            |    0.94493           |    0.94190            |\nvgg16            |    0.95665           |    0.93568            |\nvgg19            |    0.95009           |    0.93775            |\nvgg11_bn         |    0.94142           |    0.93775            |\nvgg13_bn         |    0.94423           |    0.92738            |\nvgg16_bn         |    0.94634           |    0.94190            |\nvgg19_bn         |    0.94915           |    0.93360            |\nalexnet          |    0.91260           |    0.90456            |\nresnext101_64x4d |    0.98055           |    0.96887            |\nresnext101_32x4d |    0.98172           |    0.96887            |\nnasnetalarge     |    0.96907           |    0.96265            |\ninceptionresnetv2|    0.96134           |    0.95435            |\ninceptionv4      |    0.96930           |    0.96473            |\n\n\n## Case-2 - Freezed first few layers(look at code)\nModels           | Train Accuracy_score |    Val Accuracy_score |\n-------------    | :------------------: | ----------------------:\nresnet18         |    0.9196            |    0.91493            |\nresnet34         |    0.94845           |    0.93983            |\nresnet50         |    0.9564            |    0.93983            |\nresnet101        |    0.96790           |    0.96265            |\nresnet152        |    0.96508           |    0.95643            |\nsqueezenet1_0    |    0.94048           |    0.92738            |\nsqueezenet1_1    |    0.93088           |    0.92116            |\ndensenet121      |    0.95173           |    0.95228            |\ndensenet169      |    0.96087           |    0.94813            |\ndensenet201      |    0.95384           |    0.95020            |\ninception_v3     |    0.94025           |    0.93775            |\nvgg11            |    0.93697           |    0.92946            |\nvgg13            |    0.93533           |    0.92323            |\nvgg16            |    0.94821           |    0.93983            |\nvgg19            |    0.95243           |    0.94190            |\nvgg11_bn         |    0.93416           |    0.92738            |\nvgg13_bn         |    0.93322           |    0.92323            |\nvgg16_bn         |    0.94728           |    0.93775            |\nvgg19_bn         |    0.94798           |    0.94190            |\nalexnet          |    0.89784           |    0.88589            |\nresnext101_64x4d |    0.98617           |    0.96887            |\nresnext101_32x4d |    0.98195           |    0.96473            |\nnasnetalarge     |    0.95970           |    0.96265            |\ninceptionresnetv2|    0.95103           |    0.94813            |\ninceptionv4      |    0.96251           |    0.93775            |\n\n\n# Results with Basic Agumentation\n\n## Case-1 (Freezed all layers except last one)\nModels        | Train Accuracy_score |    Val Accuracy_score |\n------------- | :------------------: | --------------------: |\nresnet18      |    0.77553           |    0.75518            |\nresnet152     |    0.82778           |    0.81535            |\nresnet101     |    0.82333           |    0.80290            |\nresnet50      |    0.79943           |    0.78630            |\nresnet34      |    0.78655           |    0.74688            |\nsqueezenet1_0 |    0.91447           |    0.87966            |\nsqueezenet1_1 |    0.90089           |    0.87344            |\ndensenet121   |    0.80880           |    0.81120            |\ndensenet169   |    0.84746           |    0.82987            |\ndensenet201   |    0.86621           |    0.86514            |\ninception_v3  |    0.76101           |    0.74688            |\nvgg11         |    0.78209           |    0.78008            |\nvgg13         |    0.75960           |    0.72821            |\nvgg16         |    0.77038           |    0.71576            |\nvgg19         |    0.71204           |    0.64522            |\nvgg11_bn      |    0.76522           |    0.74481            |\nvgg13_bn      |    0.76241           |    0.76348            |\nvgg16_bn      |    0.76265           |    0.75726            |\nvgg19_bn      |    0.75773           |    0.73858            |\nalexnet       |    0.83153           |    0.76348            |\n\n\n## Case-2 Finetuning the entire network\nModels        | Train Accuracy_score |    Val Accuracy_score |\n------------- | :------------------: | ----------------------:\nresnet18      |    0.98477           |    0.96058            |\nresnet152     |    0.99273           |    0.97717            |\nresnet101     |    0.99367           |    0.97717            |\nresnet50      |    0.99133           |    0.97510            |\nresnet34      |    0.98969           |    0.97095            |\nsqueezenet1_0 |    0.96274           |    0.94190            |\nsqueezenet1_1 |    0.96485           |    0.92738            |\ndensenet121   |    0.99086           |    0.96887            |\ndensenet169   |    0.99507           |    0.97510            |\ndensenet201   |    0.99390           |    0.97717            |\ninception_v3  |    0.98898           |    0.97302            |\nvgg11         |    0.98031           |    0.95020            |\nvgg13         |    0.98078           |    0.95643            |\nvgg16         |    0.98266           |    0.95435            |\nvgg19         |    0.98430           |    0.95643            |\nvgg11_bn      |    0.98570           |    0.96265            |\nvgg13_bn      |    0.98687           |    0.97095            |\nvgg16_bn      |    0.99179           |    0.96680            |\nvgg19_bn      |    0.99297           |    0.96680            |\nalexnet       |    0.95970           |    0.92946            |\n\n\n## Submissions:\n- densenet201 LB - to 97.22\n- ensemble1 - mode of all best performing models LB - 97.32\n\n## TO_DO\n1) Adding mixup strategy to all the networks\n2) Ensembling model outputs\n3) Model stacking\n4) Extracting bottleneck features and using\n        - ML to train the model\n        - Visualization using T-sne\n5) Solve issue with bninception(Model is not training)\n6) Train Vggm network \n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fprakashjayy%2Fpytorch_classifiers","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fprakashjayy%2Fpytorch_classifiers","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fprakashjayy%2Fpytorch_classifiers/lists"}