srihari-humbarwadi / vehicle-color-recognition-using-convolutional-neural-network

Keras implementation of https://arxiv.org/abs/1510.07391

Hi there 👋

I am a Deep Learning Software Engineer at Intel and a grad student at the University of Edinburgh.

• 🔭 I’m currently working on:
  • Implementing Panoptic Segmentation for TensorFlow Model Garden
  • Implementing Object Detection models in TensorFlow2.x
  • Self Supervision for pretraining Image models
  • Large minibatch training on TPUs
• 🌱 I’m currently learning:
  • Optimization methods, Multivariate Calculus and Machine Vision
• 👯 I’m looking to collaborate on:
  • Implementing research papers in computer vision
• 🤔 I’m looking to help others with:
  • Bugs in Python, TensorFlow1.x and 2.x code
• 💬 Ask me about:
  • Deep Learning, TensorFlow, Math
• 📫 How to reach me:
  • LinkedIn
  • Twitter
• ⚡ Fun fact: I rarely sleep at night

Recently Activity

*auto updated every hour with github actions

Last Updated at : 23:15 20-10-2023 UTC

---

Recent Activity	Details	Time
pushed 20 commits to main in srihari-humbarwadi/minimal-ml-template	* update * update * Update dockerfile * updates * ...	11:05 25-12-2022 UTC
added @AntreasAntoniou as a collaborator to srihari-humbarwadi/minimal-ml-template		17:19 12-12-2022 UTC
pushed 1 commit to dev in srihari-humbarwadi/minimal-ml-template	* Wrap example code in a function *	05:36 12-12-2022 UTC
opened a pull request in srihari-humbarwadi/minimal-ml-template	Added initial EvolutionAugment code	09:02 10-12-2022 UTC
pushed 1 commit to dev in srihari-humbarwadi/minimal-ml-template	* Initial commit *	16:42 09-12-2022 UTC
created a new branch srihari/dev in srihari-humbarwadi/minimal-ml-template		07:26 06-12-2022 UTC
pushed 5 commits to main in srihari-humbarwadi/minimal-ml-template	* update * Add remaining features for v1.0 readine...	07:25 06-12-2022 UTC
forked BayesWatch/minimal-ml-template		14:52 30-11-2022 UTC
starred BayesWatch/minimal-ml-template		14:52 30-11-2022 UTC
pushed 1 commit to main in srihari-humbarwadi/GATE	* Clone src from repo; added build arg BRANCH *	20:11 27-11-2022 UTC
pushed 5 commits to main in srihari-humbarwadi/GATE	* Added requirements.txt * Removed install_depende...	19:52 27-11-2022 UTC
pushed 1 commit to main in srihari-humbarwadi/GATE	* Create Dockerfile.cpu *	08:23 29-10-2022 UTC
pushed 1 commit to main in srihari-humbarwadi/GATE	* Update install_dependencies.sh *	08:22 29-10-2022 UTC
starred MadryLab/EditingClassifiers		16:34 28-10-2022 UTC
pushed 16 commits to main in srihari-humbarwadi/GATE	* refactor: refactoring GATE to shave off GCM and ...	06:57 21-10-2022 UTC
forked AntreasAntoniou/GATE		13:14 20-10-2022 UTC
pushed 1 commit to main in srihari-humbarwadi/miniGATE	* Create docker-compose.yml *	09:59 13-10-2022 UTC
pushed 1 commit to main in srihari-humbarwadi/miniGATE	* Update Dockerfile.cpu *	09:59 13-10-2022 UTC
pushed 1 commit to main in srihari-humbarwadi/miniGATE	* Create install_dependencies.sh *	04:42 13-10-2022 UTC
pushed 1 commit to main in srihari-humbarwadi/miniGATE	* use pytorch-cpu base image *	04:13 13-10-2022 UTC
pushed 1 commit to main in srihari-humbarwadi/miniGATE	* Added Dockerfile.cpu *	03:54 13-10-2022 UTC
forked AntreasAntoniou/miniGATE		08:20 11-10-2022 UTC
starred ggerganov/whisper.cpp		17:03 10-10-2022 UTC
starred OML-Team/open-metric-learning		16:09 09-10-2022 UTC
commented on an issue in onnx/tensorflow-onnx	@jasdeep06 I created my own quantized weight layer...	14:15 02-10-2022 UTC

Recent Commits

Message	Repository	URL	Committed At
added table entry for resnet101 model results	KamaxiB/objTens	commit/1f765c5	20:12 02-05-2022 UTC
added table entry for resnet101 model results	tensorflow/models	commit/1f765c5	20:12 02-05-2022 UTC
added table entry for resnet101 model results	sherryy88/rberry	commit/1f765c5	20:12 02-05-2022 UTC
added table entry for resnet101 model results	mhd-aldawood/framework-AI-python	commit/1f765c5	20:12 02-05-2022 UTC
added table entry for resnet50 model results	KamaxiB/objTens	commit/12ec618	02:56 29-04-2022 UTC
added table entry for resnet50 model results	tensorflow/models	commit/12ec618	02:56 29-04-2022 UTC
added table entry for resnet50 model results	sherryy88/rberry	commit/12ec618	02:56 29-04-2022 UTC
added table entry for resnet50 model results	mhd-aldawood/framework-AI-python	commit/12ec618	02:56 29-04-2022 UTC
fixed description	KamaxiB/objTens	commit/80e9b6d	02:43 29-04-2022 UTC
fixed description	tensorflow/models	commit/80e9b6d	02:43 29-04-2022 UTC

Recent Pull Requests

Title	Repository	URL	Last Updated	State
Added initial EvolutionAugment code	srihari-humbarwadi/minimal-ml-template	pull/1	17:20 12-12-2022 UTC	open
Support per sample topk loss for semantic segmentation	tensorflow/models	pull/10616	20:36 16-11-2022 UTC	open
Use ResNet v1 as default backbone	srihari-humbarwadi/models	pull/19	16:56 26-04-2022 UTC	closed
Fixes semantic loss top_k computation	srihari-humbarwadi/models	pull/18	20:00 01-04-2022 UTC	closed
Use depthwise_separable_conv2d	srihari-humbarwadi/models	pull/16	17:12 28-03-2022 UTC	closed
Added Panoptic deeplab	tensorflow/models	pull/10537	18:10 02-06-2022 UTC	closed
panoptic deeplab dataloader	srihari-humbarwadi/models	pull/15	18:50 10-03-2022 UTC	closed
panoptic deeplab modeling	srihari-humbarwadi/models	pull/14	18:49 10-03-2022 UTC	closed
panoptic-segmentation sync	srihari-humbarwadi/models	pull/13	16:47 09-03-2022 UTC	closed
Added dataloader for Panoptic Deeplab	tensorflow/models	pull/10524	16:32 16-03-2022 UTC	closed

Recent Issues

Title	Repository	URL	Last Updated	State
[TensorFlow2] Critic Loss Calculation for actor_critic	philtabor/Youtube-Code-Repository	issues/41	16:05 25-01-2022 UTC	open
INT8 calibration for efficientdet	NVIDIA/TensorRT	issues/1498	05:08 29-11-2022 UTC	closed
QuantizeLinear and DequantizeLinear nodes for Conv2D get folded as constants.	onnx/tensorflow-onnx	issues/1719	11:09 16-03-2022 UTC	closed
Allow narrow_range=1 for FakeQuantWithMinMaxVars	onnx/tensorflow-onnx	issues/1713	14:15 02-10-2022 UTC	closed
TensorRT model outputs/performance keeps changing across conversions	tensorflow/tensorflow	issues/51719	21:18 11-10-2021 UTC	open
Support resizing images on TPU using tf.image.resize when size is not a compile-time constant	tensorflow/tensorflow	issues/51693	18:02 18-11-2022 UTC	closed
Adding support for reading and writing to multiple tfrecords in nsl.tools.pack_nbrs	tensorflow/neural-structured-learning	issues/92	19:06 12-01-2022 UTC	open
Not able to capture profile from Tensorboard profile tab	tensorflow/tensorboard	issues/4954	17:10 11-05-2021 UTC	closed
Misleading warning when loading weights from hdf5 format weights	tensorflow/tensorflow	issues/46591	22:03 24-03-2022 UTC	open
TensorRT converter fails for CombinedNonMaxSuppression	tensorflow/tensorflow	issues/46453	00:56 12-03-2021 UTC	closed

---

vehicle-color-recognition-using-convolutional-neural-network's Issues

About training Codes

Hi, Have you already trained on dataset? Thanks for your excellent job. @srihari-humbarwadi

how to run this code?

Hi, I need your help.
how to run this code? and how to predict with this code?
Or anyone knows, please tell me! Many thanks!

Number of epochs

Your implementation use 50 epochs. The paper says that they used 200000 iterations. The training set is composed of 7799 images. Using a batch size of 32 this would be around 820 epochs.

Optimization

Your implementation is using Adam when in the paper is said that they use stochastic gradient descent optimization.

Wrong Normalization

Your implementation uses Batch Normalization but in the paper is said that the network use Local Response Normalization.

can you release the .h5 file?

Hi,
in this line
chkpt = ModelCheckpoint('models/top_weights.h5', monitor='val_acc', save_best_only=True, save_weights_only=True,verbose=1)

have 'models/top_weights.h5'
but it not found in the github repo

can you release this file?
many thanks!

The .h5 file

In batch_training.py, I can see a .h5 file which is weight of pretrained model. So can you release it? and What is it?

Thank you