Mass processing of product photos using AI

Budget: 11000 UAH Deadline: 4 days

Hello! The task is clear: you need to batch process 38,000 photos from Google Drive using AI (detection of two types of photo zones, Inpainting/Inpainting-lama with the removal of brand elements and background restoration without touching the product, generation of updated structure and Excel).

I will implement this in Python:

Asynchronous upload and download of photos via Google Drive API / PyDrive2.

AI pipeline for generating masks of photo zones and their inpainting (ControlNet / LaMa / Inpainting models) with processing on GPU based on automation scripts.

Validator for automatic verification filtering of photos without photo zones (remain unchanged).

Automatic creation of folder structure and generation of final Excel (openpyxl).

I am ready to do a free test on 10-20 photos from your folder to agree on quality. Please write in private messages, and we will discuss the details and provide an exact estimate. Expand

Budget: 26460 UAH Deadline: 7 days

Hello, I recently worked on a similar mass photo processing pipeline for e-commerce: automatic detection of photo zones, removal of branded elements (logos, phone numbers, text) while restoring the background based on texture and lighting, keeping the product unchanged. Reading Excel with ad numbers and URLs, navigating the folder structure on Google Drive, and saving a new tree with the same numbers was also part of the same task.

If you have photos with two photo zones in one frame, it’s worth adding a step for detecting multiple zones and queuing for manual review of questionable cases, which reduces defects by about 10-15%.

Is this a one-time processing or will there be regular batches, and do you have examples of photos where the photo zone occupies a larger part of the frame, behind the product? This affects the choice of inpainting model.

I suggest we get in touch, I will prepare a pipeline scheme for your 38,000 photos in Python with zone detection and inpainting, along with timelines and technologies tailored to your case.

Budget: 11000 UAH Deadline: 5 days

38,000 photos is the scale where the question is not "will AI remove the logo," but "what will happen in those two percent where it makes a mistake." 2% of 38,000 is 760 spoiled photos, and finding them visually is already impossible. Therefore, I am building the pipeline around control, not around the inpainting itself.

How I will do it:
1. Calibration on 300–500 photos from different ads and angles. I will show the result and we will fix the acceptance criteria: what we consider a clean background, what is acceptable, and what is not. Only after your "yes" will I launch the rest.
2. Detection of the photo zone is a separate step from removal. I classify each photo: zone #1 / zone #2 / no zone, and save the decision with a confidence rating. I do not touch photos without a zone at all, as you requested.
3. Removal of brand elements through inpainting using masks for the specific zone, restoring the texture, color, and lighting of the background. I protect the edges of the product with a separate mask so that the edges are not "eaten away."
4. Automatic quality control: photos with low detection confidence or with an abnormally large area of changes do not go to "ready," but to a separate folder "for review." This is the main difference from "batch processing" — you see the questionable cases, rather than receiving them quietly among the rest.
5. At the output: a new folder on Google Drive with the same structure by ad numbers, an updated Excel table with links, and a third file — a report: how many have been processed, how many have been left unchanged, how many are "for review," and why.

Technologies: Python, Google Drive API, zone detection + inpainting (LaMa or SD-inpaint — I will choose after calibration, as it is visible on your real photos which holds the background better), pandas for the table.

11,000 UAH, 5 days including calibration.

Question: Is the photo zone a physical banner or wall with writings behind the product, or are the writings and logos overlaid through post-processing on top of the frame? These are different risk tasks: overlaid elements are removed almost without traces, while a physical banner has to be redrawn — and this is where the percentage of defects comes from, for which calibration is needed. Expand

Budget: 20000 UAH Deadline: 21 days

Hello! The task is clear: you need to process approximately 38,000 photos, identify photo zones, and remove branded elements (logos, phone numbers, text) without touching the product and background.

For this, I would use a combination: detection of photo zones through a vision model (GPT-4o or Google Vision), inpainting via Stable Diffusion with ControlNet or SDXL Inpaint, and coordination of the entire process through a Python script that reads Excel, takes URLs/folders from Google Drive, and saves the result in a new structure with an updated table.

The critical point here is the quality of inpainting: the background needs to be preserved naturally, so before starting, I will conduct a test on approximately 100-200 photos and show the result for confirmation. Processing in batches using Google Drive API, without manual uploads.

Estimated timeframe: 2-3 weeks after confirming the quality of the test. Cost: $600-900 depending on the complexity of the photo zones and the number of iterations. I am ready to discuss the details.

Budget: 7900 UAH Deadline: 10 days

Hello! I am Serhiy, I have 8 years of experience in IT across various technologies: web/mobile, backend, API, integrations, automation, and data handling.
I see the main task as follows: Mass processing of product photos using AI. A one-time mass processing of product photos using AI is required. The output data will be an Excel spreadsheet with...

Skills for the task:
- AI/ML, RAG, and automation
- solution architecture
- development, testing, and release
- post-launch support

Portfolio: https://software.campstudio.net/#showcases

Budget: 4000 UAH Deadline: 10 days

Good day!

I understand why this task is difficult to estimate in advance: 38,000 photos from different angles and distances is not typical "batch" processing; the quality of recognizing the photo zone in atypical shots is crucial here.

Therefore, I suggest not just taking my word for it but checking it in practice: I am ready to process 5-10 of your actual photos for free (different angles, both options of the photo zone) within a day after gaining access—so you can see the results before any payment or agreement.

Technologies: photo zone detection (considering that there are only two templates—this simplifies training/setup) + inpainting model (LaMa) for background restoration without traces of the logo, preserving lighting and texture. The pipeline is fully automated: Excel → Google Drive API → processing → new folder structure by ad numbers → updated Excel with links.

Plan: free test (today-tomorrow) → if the quality is satisfactory → pilot on 100-200 photos with final accuracy fixation → full volume of 38,000 photos, approximately 10-15 working days.

Cost: test—free. I will evaluate the full volume accurately after the test (at a progressive rate of $X/1000 photos—less for simple shots, more for complex ones).

I am ready to start today—please send a link to a couple of folders with different types of photos for the test. Expand

Budget: 4000 UAH Deadline: 7 days

Good day
I can take on your task. I will take each photo and process it through AI (OpenAI) using a prompt to clean it from unnecessary elements. In the end, you will receive a folder in Google Drive with the photos and a new Excel table with updated links. The cost for the work is 4000 UAH for writing scripts + the cost of using artificial intelligence. The cost of AI work can be estimated after we process the first batch of photos and agree on the quality of the final images.

Budget: 7200 UAH Deadline: 30 days

We already have an almost ready module for mass AI processing of product photos - it can be quickly adapted and launched, I suggest discussing the details here, I am available ))

- total cost - 7200 USD
- execution time - 30 calendar days
- technologies - Python, OpenCV, PyTorch, product and photo zone segmentation, product protection masks, controlled background restoration, GPU processing, Google Drive API, and OpenPyXL

The cost includes a pilot sample, setup of two photo zones, processing of 38,000 photos, result verification, reprocessing of deviations, maintaining Google Drive structure, and updating Excel

- Are there benchmark photos of both photo zones without branding and products?
- Do you agree to a pilot of 200-300 complex photos as a criterion for launching the entire array?

https://business.ingello.com/ua/vorfahr - AI processing and enhancement of product photos

https://business.ingello.com/ua/prime-eva - automation of products, production, and integrations

https://systems-fl.ingello.com/ua - team and work format Expand

Budget: 9000 UAH Deadline: 5 days

Hello, I can perform a one-time mass AI processing of product photos according to an Excel spreadsheet and folder structure on Google Drive. I will create a Python script that matches the ad numbers with the photos, processes the images according to the specified rules, and saves the result in the required structure. Before the mass launch, I suggest agreeing on 5-10 test examples to ensure the quality of processing and avoid manual revisions. I have worked with data processing automation, images, APIs, and batch tasks, so I can do it carefully and with result verification. git: github.com/onyx144

Budget: 3400 UAH Deadline: 4 days

Task: detection of the photo zone, removal of logos and inscriptions using inpainting model, background and lighting unchanged.

Processing 38,000 photos, I will keep the folder structure on Drive, Excel will be updated with new links.

Deadline approximately 4 days.